Dignaga on the Interpretation of Signs (Studies of Classical India; v. 9)

Dignaga on the Interpretation of Signs

Studies of Classical India Editors Bimal K. Matilal Spalding Professor of Eastern Religions & Ethics, Oxford University, England

J. Moussaieff Masson Professor of Sanskrit, University of California, j]erkeley, U.S.A.

Editorial Board: R. P. Goldman, Daniel H. H. Ingalls, and A. K. Ramanujan

The aim of this series is to publish fundamental studies concerning classical Indian civilization. It will include editions of texts, translations, specialized studies, and scholarly works of more general interest related to various fields of classical Indian culture such as philosophy, grammar, literature, religion, art, and history. In this context, the term 'Classical India', covers a vast area both historically and geographically, and embraces various religions and philosophical traditions, such as Buddhism, Jainism, and Hinduism, and many languages from Vedic and Epic Sanskrit to Pali, Prakrit, and Apabhraqula. We believe that in a profoundly traditional society like India, the study of classical culture is always relevant and important. Classical India presents an interesting record of deep human experience, thoughts, beliefs, and myths, which have been a source of inspiration for countless generations. We are persuaded of its lasting value and relevance to modern man. By using extensive and for the most part unexplored material with scientific rigor and modern methodology, the authors and editors of this series hope to stimulate and promote interest and research in a field that needs to be placed in its proper perspective.

Volume 9

Dignagaon the Interpretation of Signs

by

Richard P. Hayes /'

Faculty o/Religious Studies, McGill University, Montreal, Quebec, Canada

KLUWER ACADEMIC PUBLISHERS DORDRECHT I BOSTON I LONDON

Library of Congress Cataloging in Publication Data Hayes, Richard P. (Richard Philip), 1945Dignaga on the interpretation of signs. (Studies of classical India; v. 9) Composite revision of three earlier works, one of which was the author's thesis (Ph. D.-University of Toronto). Includes translation of Ch. 2 & Ch. 5 of Pramanasamuccaya. Bibliography: p. Includes index. 1. Dignaga, 5th cent. 2. Languages-Philosophy. 3. Languages-Religious aspects-Buddhism. 1. Digniiga, 5th cent. Pramanasamuccaya. Ch. 2. English. 1987. II. Dignaga, 5th cent. Pramanasamuccaya. Ch. 5. English. 1987. III. Title. IV. Series. 181' .043 87-35620 B133.D654H39 1987 ISBN 90--277-2667-1

Published by Kluwer Academic Publishers, P.O. Box 17, 3300 AA Dordrecht, Holland. Kluwer Academic Publishers incorporates the publishing programmes of D. Reidel, Martinus Nijhoff, Dr W. Junk and MTP Press. Sold and distributed in the U.S.A. and Canada by Kluwer Academic Publishers, 101 Philip Drive, Norwell, MA 02061, U.S.A.

In all other countries, sold and distributed by Kluwer Academic Publishers Group, P.O. Box 322, 3300 AH Dordrecht, Holland.

All Rights Reserved

© 1988 by Kluwer Academic Publishers No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner Printed in The Netherlands

To Philip T. and Helen Schooley Hayes, who by lifelong example have taught the author to treasure critical thinking, to shun dogmatism, and to sUstain good humour in a world that often disappoints.

CONTENTS Preface . . . . • . . • Acknowleagements • . .

Chapter 1: Preliminaries 1.1 The central issues ••. •. 1.2 The contributions of previous scholars 1.2.1 Satis Chandra VidyabhU$aI)a . . . . 1.2.2 Th. Stcherbatsky . . . . . . . . . 1.2.3 Satkari Mookerjee . . . . . . . . 1.2.4 Erich Frauwallner . . . . . . . . . . . . . . . 1.2.5 Kitagawa Hidenori, Hattori Masaaki and Katsura Shoryu 1.2.6 Radhika Herzberger . .. . . . . . . . . . . 1.3 The argument of this book • . • • • • • . • • • . Notes . . . . . . . . . . . . . . . . . . . . . . . Chapter 2: Rational Skepticism in Pre-Diimagan Buddhism 2.1 The foundation of skepticism in the Nikiiyas ••• 2.1.1 The Sutta Nipiita . . . . . . . . . . . . . . . 2.1.2 DIgha Nikaya: The Brahmajlila sutta . . . . . . . 2.1.3 AJiguttara Nikaya: The Kesaputtas . . . . . . . . 2.1.4 Summary of how opinions are regarded in the Nikayas 2.2 The influence of Nagarjuna

2.2.1 Malamadhyamakakarika . . . . . ... 2.2.2 VigrahavyavartanI . . . . . . . . . . . Notes . . . . . . . . . . . . . . . . . . . Chapter 3: Nominalism in Pre-Ditmagan Buddhism 3.1 The Agama literature and Milindapaiiha • • 3.1.1 Natural class in the Nikayas 3.1.2 Personal identity in the Piili Canon . . 3.1.3 Personal identity in the Milindapaiiha 3.2 Nominalism in Nagarjuna. • • 3.3 Nominalism in Vasubandhu •• 3.3.1 Vasubandhu's theory of two truths 3.3.2 Vasubandhu's phenomenalism . Notes . . . . . . . . . . . . . . Chapter 4: Dinnaga's Theory of Knowledge

4.1 Hetucakranirl}aya • • • • • • • 4.2 The Pramii1J.asamuccaya • • • • • 4.2.1 Sensation in thePramii1J.asamuccaya 4.2.1.1 Awareness's awareness of itself . 4.2.2 Inference in the PramiiT)asamuccaya 4.2.2.1 The subject matter of inference . . . . 4.2.2.2 Three characteristics of legitimate evidence 4.2.2.3 On errancy and pervasion . . . . . . . . . 4.3 The skepticism implicit in Dmnaga's epistemology Notes . . . . . . . . . . . . . . . . . . . . .

ix xv

· 1 · 1 • 9

· 9

.11 .16 .21 .24 .30 .32

.36 .41 .42 .43 .45 .48 .50 .52 .53

.59 .63

.72 .78 .79 .81 .86 .88 .93 .93

.96 104

111 111 131 133

140 142 143 145 154

158 168

viii

Contents

Chapter 5: Diilnliga's nominalism ...............o'.. 5.1 The Alambanaparfk$ii . • . • . • • • • . • • • • • • • • . . . 5.2 The context of the discussion of nominalism in the Pramiiq,asamuccaya 5.2.1 Scripture as a form of inferentiall)ign . 5.2.2 Fallibility in inference and scripture . . . . . 5.2.3 The question of universals . . . . . . . . . 5.2.4 AnyClpoha as a substitute for universals . . . . 5.2.5 The nature of information conveyed by language 5.2.6 The meaning of individual words . . . . . . 5.2.7 Particulars as instantiations of universals . . . . 5.2.8 Absurdities in the view that universals exist outside thought 5.2.9 The contrariety of expressions . . . . . . 5.2.10 The meaning of a sentence . . . . . . . 5.2.11 The sentence as the primary linguistic symbol Notes. . . . . . . . . . . . . . . . . . .

173 173 178 178 179 183 185 188 193 196 204 205 212 215 216

TRANSLATIONS Introduction to translation . . . . . . . . . . The history of the PramClTJaSamuccaya in Tibet . .

223 226

Chapter 6: Pramiiq,asamuccaya n "On reasoning" 6.1 Inference differentiated from sensation . 6.2 The three characteristics of legitimate evidence 6.3 Property-bearer as the subject of inference 6.4 On restricted and errant properties . . . . 6.5 Non-symmetry of restriction and pervasion • . Notes. . . . . . . . . . . . . . . . . . . .

231 231 239 242 244 247 249

Chapter 7: Prama"asamuccaya V: On the nature of signs in language 7.1 On the question of what verbal symbols make known 7.2 On tbe relationships between symbols that express preclusion 7.3 On the unreality of universals outside thought 7.4 On the question of what linguistic symbols preclude Notes. . . . . . .

252 252 278 282 287 300

Chapter 8: Conclusions • . . • . . .

309

Appendix A: Glossary of Sanskrit Terms Appendix B: Tibetan-Sanskrit Lexicon . . . . . . . . Appendix C: Text key to translations of PramClr;asamuccaya Selected Bibliography . . . . . . . . . . . . . . . . Subject and Author Index . . . . . . . . . . . . . . .

313 322 332 335 353

Preface

Buddhist philosophy in India in the early sixth century c.E. took an important tum away from the traditional methods of explaining and systematizing the teachings in Sutra literature that were attributed to the Buddha. The new direction in which several Indian Buddhist philosophers began to move was that of following reasoning to its natural conclusions, regardless whether the conclusions conflicted with traditional teachings. The central figure in this new movement was Diimaga, a native of South India who found his way to the centre of Buddhist education at Nalanda, studied the treatises that were learned by the Buddhist intellectuals of his day, and eventually wrote works of his own that formed the core of a distinctly new school of Buddhist thought. Inasmuch as virtually every Indian philosopher after the sixth century had either to reject Diimaga's methods or build upon the foundations provided by his investigations into logic, epistemology and language, his influence on the evolution of Indian philosophy was considerable, and indeed some familiarity with Dinnaga's arguments and conclusions is indispensable for anyone who wishes to understand the historical development of Indian thought. Moreover, since the approach to Buddhism that grew out of Dinnaga's meditations on language and the limits of knowledge dominated the minds of many of the scholars who took Buddhism to Tibet, some familiarity with Dinnaga is also essential to those who wish to understand the intellectual infrastructure of Tibetan Buddhist philosophy and practice. Despite Dinnaga's importance in the evolution of Indian and Tibetan philosophy, however, it has not been easy for modem scholars to gain access to his ideas. All his key works have perished in their original Sanskrit versions and can be read only in rather awkward Tibetan and Chinese translations. But because so many of Dinnaga's observations, especially of language, were deeply influenced by the Sanskrit grammatical tradition of Pfu).ini, Patafijali and Bhartrhari, his philosophical writings were extremely difficult to translate into comprehensible Tibetan or Chinese, so even these translations can hardly be read. Rather, they must be painstakingly deciphered by a very slow and complex procedure that involves piecing together clues that have been gathered not only from the Buddhist and Brahmanical philosophical traditions but also from India's rich grammatical tradition. At times the work feels more like the task of an archaeologist fitting together the scattered fragments of an unearthed pot than like the task of an historian of philosophy, for a great

ix

x

Preface

deal of effort at the outset must be expended on discovering' even what Diimiiga had said, not to think of what significance might have been attached to saying it. But gradually the pieces begin to fit together, and the result of the labours of piecing them together has been the present book, which focuses on Diimiiga's theories on the interpretations of natural signs in ordinary inference and of conventionally contrived signs in human discourse. My initial intention, when I began working on this general topic in 1974, was to gain some understanding of the theory of language that featured so prominently in the works of such Indian Buddhist writers as DharmakIrti, Dharmottara, Siintarak~ita, Kamalasila, Ratnakirti and Jiiiinasnmitra. The decision to do research for a doctoral dissertation in this area had come about as a result of feeling torn between three loves: Buddhist philosophy, formal logic and Sanskrit grammar. Studying the linguistic theory of the Buddhist epistemologists seemed an ideal way to continue research in all three of these areas, and I shall always be grateful to Professors Katsura Shoryu and B.K. Matilal for pointing me towards a study of the linguistic philosophy of Santarak~ita for my dissertation research. The more I worked on Santarak~ita, however, the more apparent it became that I must first gain a clear understanding of the antecedents to his thinking about language, and so began the study of numerous thinkers whose criticisms of the Buddhist philosophy of language Santarak~ita had tried to answer. Ultimately, of course, it was necessary to come to terms with the principal source of most of the ideas developed by Buddhist epistemologists and linguists, and so I came to the study of Diimiiga's Pramiil)asamuccaya.

As I began to work through the section of the Pramiil)asamuccaya that deals especially with philosophy of language, I expected to find there a series of arguments and conclusions essentially like those that are to be found in the work of Dinniiga's most celebrated interpreter, Dharmakirti. What was in fact to be found in reading Dinniiga at first hand, however, was rather different from accounts found in virtually all the secondary literature that treats Buddhist linguistic philosophy, and I came to form the unexpected conclusion that nearly ~very modem scholar of classical Buddhist thought has proceeded on the false assumption that Dmniiga and Dharmakirti had advanced essentially the same arguments towards essentially the same conclusions. It was with great delight that I discovered in Dinniiga not merely an inchoate and simpler version of Dharmaldrti, but a strikingly distinct thinker whose agenda was different from DharmakIrti's in a number of important ways. The ways in which Dinuiiga differed from his later interpreters are spelled out in the introduction and

Preface

xi

conclusion of this book. Stated simply, the difference is this: whereas DharmakIrti was the architect of a complex edifice of apologetics in which every received dogma of Indian Buddhism was justified by a multiplicity of arguments and every cherished Brahmanical belief was subjected to a barrage of feisty polemics, DiIinaga emerged as a figure much more in line with the skeptical spirit of archaic Buddhism and early MMhyamaka philosophy. For DiIinaga the central task was not to construct and defend a rationalized system of thought but to examine the fundamental assumptions on which all our claims to understanding rest. DiIinaga's principal philosophical work was called the PramiifJ,asamuccaya, for it was a collection (samuccaya) of shorter writings that he had done on various aspects of the study of what constitutes knowledge (pramiifJ,a). The present work is also a samuccaya of sorts in that it

comprises a number of works that were originally independent studies of aspects of DiIinaga's thought. An observation that stands at the heart of DiIinaga's investigation is that language conveys information of essentially the same type that inference conveys. That is, upon hearing a spoken sentence the hearer forms an hypothesis about the idea that was in the mind of the speaker. A sentence is the sign of an idea in about the same way that a body of smoke is a sign of fire. Interpreting the sign of a thought expressed by a sentence in a conventional human language is governed by the same constraints and carries with it the same risks of error as does drawing an inference from any sign in nature. Understanding what the risks of error are in understanding a sign that takes the form of a sentence requires first understanding what the risks are of drawing conclusions from observed signs in general. Given the intimate connection between the theory of inference and the theory of language in DiIinaga' s philosophy, it makes sense to treat these two branches of DiIinaga's inquiry together in a single volume. For this reason I have merged together, with a considerable amount of editing and rewriting, three of my earlier studies of DiIinaga's philosophy. Two of these earlier studies were in the form of articles (Hayes 1980 and 1984), and the third was my Ph.D. dissertation at University of Toronto. The 1980 article contained a translation of the main part of Pramii1).asamuccaya chapter two, which deals with inference, and the dissertation contained a study and translation of Pramii1).asamuccaya chapter five, which deals with linguistic signs. The 1984 article dealt with aspects of DiIinaga's theory of inference and tied his theory of inference to his theory of language. In the present work I have tied together somewhat modified and updated forms of these three earlier works.

xii

Preface

But during the time that I was working on Dirinaga's 'thought, it happened that I was also teaching courses in Indian philosophy, in Buddhist philosophy and in Buddhist religious practice at the University of Toronto, and my thought turned more and more to seeing how Dirinaga's philosophical writings fit into the overall development of not only Buddhist theory but also Buddhist practice. The more I reflected on Buddhism in general and Dirinaga in particular, the more it struck me that a careful study of the nature of language was not merely an interesting intellectual hobby to a man who happened to be a Buddhist monk but rather that exploring the limits of communication and the limits of human understanding were central to the task of achieving nirvcl1:ta as traditionally understood in mainstream Buddhism. Therefore, while the focus of my dissertation was on Dmnaga's philosophy of language as such, the present book deals much more fully with the question of why a Buddhist qua Buddhist would bother to have a philosophy of language at all. It is my contention that Dirinaga's theory of inference is essentially skeptical and that his theory of language is also skeptical. It is also my contention that skepticism, empiricism and nominalism go very comfortably together as philosophical views and that it is therefore no surprise that Dirinaga the skeptic should also be an empiricist and a nominalist. Finally, it is my contention that skeptical and nominalistic tendencies are present in Buddhism from the very outset and so there is no great surprise in finding that a Buddhist should be a nominalist, a skeptic and an empiricist. But I have discovered in talking to other scholars that while I find it perfectly natural that a Buddhist would easily be a skeptic, a nominalist and an empiricist, not everyone sees the natural congruence of these different positions. Therefore I have thought it best to spell out my reasoning on this matter in some detail in chapters two and three of this study, which deal respectively with the skeptical trend and the tendency towards nominalism in canonical Buddhism, in Buddhist abhidharma and in early Madhyamaka thought as set down by Nligarjuna. At the risk of oversimplifying .the argument of my book, one might say that my conclusion is that Dirinaga, rather than being the founder of a radically new school of Buddhism, was more a natural continuation of a line of a very gradual evolution, an evolution characterized not by dramatic changes of perspective but by a natural process of discovering ever more fully the complex implications of a perspective that had at first seemed much more simple. Thus, while I doubt that the Buddha could ever have anticipated that a philosophy of language like Dirinaga's would evolve out of something as simple as the Four Noble Truths, I am also inclined to feel that a philosophy of language like Dirinaga's grows quite naturally out of the basic suppositions in the Buddha's teachings. Given both the teachings of mainstream Buddhism and the general Indian obsession with questions

Preface

xiii

of language, it is almost inevitable that a Buddhist would come along with views on language.like Diimaga's. It has been my intention in putting together this book to provide material that would be of interest to several types of reader. Thus I have ventured into areas, such as a comparison of the Hellenistic Skeptics with some trends of early Buddhism, that I hope will be of interest to historians of philosophy who are interested in comparing the evolution of thought in Europe with the evolution of thought in India. I have also dealt with topics that are intended to be of interest to historians of Buddhism and scholars engaged in the comparative studies of religion. In addition, some discussions have been included of points that scholars of the Sanskrit and Tibetan languages should find interesting. Finally, it is my hope that members of the Buddhist community, especially those in the West where Buddhism has yet to reach its mature form, will find something of value in the reflections offered here of the intellectual achievements made during one of the most creative and vital periods in the history of the religion. Very few academic books represent the work and ideas of only one scholar, although it is customary for one or occasionally two authors to take responsibility for the final form of the work. The present book would quite simply never have existed had it not been for the kind influences of a number of mentors and colleagues. As mentioned earlier, the initial suggestion that I do some work in the area of the linguistic philosophy of the Buddhist epistemologists came from Prof. Katsura Shoryu, who gave me my first instruction in 1973 at University of Toronto both in the classical Tibetan language and in the Buddhist epistemologists. My interest in the thought of DharmakIrti and his school was kept alive by Professors A.K. Warder and B.K. Matilai. As my decision to work.on the Buddhist theory of language grew more firm and it became apparent that it would be necessary to become familiar with the Brahmanical theories of language in opposition to which the Buddhist theories were formed, Prof. Matilal provided able and amiable guidance through relevant passages in Uddyotakara's Nyiiyaviirttika, Kumarila Bhatta's MfmiirhsiiSlokaviirttika, and Mallavadin's Dviidasiirarh Nayacakram. The study of these texts with Prof. Matilal provided a most valuable introduction to the issues involved in philosophy of language between the time of Dirinaga and his earliest apologist, DharmakIrti, and without this background many passages in Dirinaga's work would have been far less intelligible to me than they in fact were. At the same time Prof. T. Venkatiicarya provided instruction in the fundamentals of PiiI)ini's grammatical system, which repeatedly proved invaluable in reaching an understanding of the philosophies of Bhartrhari and Dirinaga. The study of the Tibetan translations of Dirinaga' s Pramii~LQ-

XIV

Preface

samuccaya and Jinendrabuddhi's ViHlliimalavatf PramiilJasamuccayatfkii

was carried out in the Department of. Indian Philosophy at Hiroshima University, where Prof. Katsura read through the texts with me and offered many valuable criticisms of my interpretations and English translations. His impressive command of the subject matter combined with his constantly warm-hearted encouragement made working with him among the most valued experiences of my life. Upon my return to Toronto after sixteen months in Japan, Prof. Matilal provided careful supervision of my Ph.D. dissertation and immediately after that task was completed he began to urge me to expand that work into a more comprehensive study of Diilnaga's philosophy. Thanks to comments from him and from Prof. D. Seyfort Ruegg, a number of improvements over the dissertation have been incorporated into the present work. It is with a feeling of deep gratitude to all the aforementioned professors that I put this book into the public eye, urging whoever may read it to be mindful of the fact that credit for everything of value in this book is due to these excellent teachers, who cannot, however, be held responsible for any carelessness or faulty thinking on my part. My good fortune has not been limited to having fine teachers, for I have also benefited from excellent companions. My valued friend and colleague Brendan Gillon has shaped my thinking in numerous ways through the past dozen years of collaboration, conversation and correspondence, and many parts· of this book bear the stamp of his influence. My wife Beth has not only provided moral support through the hectic and often discouraging pace of the past several years, but as a skilled Sanskritist, cultural historian and philosopher in her own right she has always been an intelligent and well-informed audience on whom to try out my ideas before committing them to the unforgiving medium of print. And when the time did come to deal with the problems specific to getting manuscript into print, Jason Weiss generously gave countless hours of valuable advice and computer programming expertise towards helping find elegant ways to make software and printers deal with Sanskrit diacritical marks. I am much indebted to him not only for his technical help but for his warm and cheerful friendship. Finally, Ven. Samu Sunim, Zen master at Zen Buddhist Temple Toronto, has put his keen intellect and sense of humour to the good use of showing me repeatedly that there is much more richness in Buddhism than just the philosophical musings that so rouse my curiosity.

Acknowledgements

The research out of which this book grew would not have been possible without generous grants from the Japanese Ministry of Education (MonbushO), which financed my period of research at Hiroshima Daigaku, and the National Endowment for the Humanities, an agency of the United States Federal Government, which financed two further years of research into Dmnaga's PramiilJasamuccaya. The expenses of hardware and software used in producing the final manuscript were defrayed by a gift from Elinor G. Hayes. Prof. Milton Israel, director of the Centre of South Asian Studies at University of Toronto, kindly arranged for me to use the University of Toronto research library and other facilities during the 1986-87 academic year. The printing of the final manuscript was done on a laser printer put at my disposal by the Zen Buddhist Temple Toronto.

xv

Chapter 1 Preliminaries 1.1 The central issues Diimaga was an Indian Buddhist philosopher who most probably wrote in the first half of the sixth century of the Common Era and who devoted a great deal of his literary career to investigating various issues in epistemology, the science devoted to the nature of knowledge and its limits. As a specific part of this larger concern, Dinnaga (whose name is frequently encountered in its alternative spelling of Dignaga) was interested in the nature and limitations of information that is gained through the interpretation of signs. The interpretation of natural signs, that is, of properties such as smoke that can be understood as signifying the presence of fIre, was investigated under the general heading of inference or reasoning (anumiina), or more specifically, reasoning for oneself (sviirthiinumiina). The interpretation of conventional signs, that is, of the words and sentences of human language and other systems of deliberately contrived symbols whose usage is governed by human-made conventions, was investigated separately as a subject unto itself, although Dinnaga explicitly regarded the interpretation of deliberately contrived symbols as no more than a special application of the principles of inference in general. The principal purpose of this book is to come to an understanding of Dmnaga's views on the interpretation of signs, both natural and conventional. The study of a philosopher from the relatively remote past can be undertaken in anyone of a number of ways. It is possible, for example, to isolate certain issues dealt with by the thinker and to treat those issues in approximately the way that such issues are discussed by modern philosophers, paying only minimal attention to the importance of those issues for the original philosopher. Proceeding in this way one might, for example, wish to focus one's attention on the geometrical theorems of Pythagoras and prescind from the issue of the place that geometry held within the overall complex of philosophical and religious beliefs and 1

2

Preliminaries

practices that Pythagoras taught. Or, to give an instance from Indian thought, one might focus on the linguistic observations of PaI)ini and ignore his religious motivations for making those observations in the first place. In a sense this approach is entirely justifiable, for to the linguist qua linguist engaged in research aimed at understanding how Pal).ini regarded certain phenomena of the Sanslait language, it makes very little difference that Pal).ini would surely never have made those observations had he not been concerned in preserving the purity of a ritual language that was sufficiently archaic that his contemporary priests could no longer use it with natural ease and confidence. Few modern linguists care much about performing a Vedic ritual properly, yet they find Pal).ini's linguistic observations an object of great interest in themselves. Alternatively it is possible in dealing with a philosopher from a former age to try to determine what significance a given range of topics had for him in his own historical context, paying relatively little attention to how or whether his treatment of those ideas appeals to us in our own place in history. Like the ahistorical philosophical approach, this historical approach to the study of human ideas has its advantages. To appreciate what a theory meant to its original author is arguably to understand that theory better than to understand simply what it means to us, and even if we grant that we can never fully understand another person's mind, it is still important to set the goal of understanding what a given theory meant to its originator. By understanding a theory in its original context we are probably in a much better position to see just how and why our own counterparts or versions of that theory differ and why the original theory has come to be modified, refined or abandoned altogether. In the present work my aim 'is to discover as well as I am able the place that the investigation 0-;' the interpretation of signs held in Diimaga's system of philosophy. My approach, therefore, is primarily historical. But throughout my work I try to avoid treating past philosophies as mere specimens on display in some intellectual museum full of quaint exhibits of thankfully outmoded world views. The basic motivation of philosophy, namely, the search for wisdom and truths by which we might lead worthwhile lives, will never become a thing of the past until intelligent life itself ceases to be a feature of the universe. And given that basic motivation, the issues and problems of philosophy are bound to remain more or less constant. What changes from one age to another is what the majority of intellectuals, or whoever else may be influencing the way the general population thinks, find acceptable as answers to those basic philosophical problems. As an historian of philosophy, of cqurse, my main interest is not whether we of this century find Dinnaga's philosophical

Chapter One

3

conclusions acceptable to our current tastes, for that is the preoccupation of an apologist. I c.an say that I would be very surprised if more than a handful of people from this century would find Diilllaga's conclusions completely satisfactory and his arguments for those conclusions convincing, simply because he was something of a pioneer, and like all pioneers his tools were rather crude and unsophisticated compared to those of the later generations who benefited by inheriting the fruits of the pioneer's labours. But there is, nevertheless, an intrinsic value in constantly reexamining the thoughts of our predecessors with a fresh mind, without assuming like the reactionary that past ages had a better handle on reality than we have and without assuming like the uncritical modernist that humanity is incessantly progressing towards some higher understanding. We can only become richer in understanding by learning from mistakes, whether they be our own or someone else's. My main interest, then, is simply to present as clearly as possible what my reading of Diilllaga has led me to believe he intended to achieve for his own contemporary readers by writing down his line of thinking about the interpretation of signs and language. The question of what place the study of epistemology, logic and language had in the overall system of Diilllaga and his followers is one to which there has been no uniform answer by modern scholars of classical Indian Buddhism. Everyone agrees that Diilllaga was a Buddhist and wrote his philosophical treatises as a Buddhist, and nearly everyone agrees that the principal and perhaps even the only interest of a Buddhist as a Buddhist is nirvli1)a, whether in gaining it for oneself in this life or in helping all beings attain it eventually. The question, then, is what the study of epistemology, logic and language has to do with nirvli1)a. Stcherbatsky, writing in the early 1930's, reports that this question is not at all new but was mooted by Indian and Tibetan followers of the DiIinaga school. According to Stcherbatsky, there were three schools of interpretation of the works of DiIinaga and his most prominent interpreter DharmakIrti.! One of the schools, says Stcherbatsky, contented itself with merely explicating the texts without exploring the religious or philosophical implications, whereas the two other schools placed a great deal of emphasis on the religious value of studying epistemology. Of these two other schools, one was supposedly dedicated to the Mahayana dogma of "the Buddha as a personification of Absolute Existence and Absolute knowledge ... a metaphysical entity, and therefore uncognizable for us, neither in the way of an affirmation nor in the way of a denial,"2 whereas the other was made up of lay followers. who practised tantric rites. In contrast to these Indian interpreters, says Stcherbatsky, there were such Tibetans as Kun dga' rgyal mtshan the Sa skya PaI)Qita who "maintained that logic is an utterly

4

Preliminaries

profane science, containing nothing Buddhistic at all, just as'medicine or mathematics are. The celebrated historian Bu-ston Rin-poche shares in the same opinion. ,,3 On the other hand, the Tibetan Dge lugs pa school is said to regard the study of logic as central to the task of founding "a sure foundation of Buddhism as a religion."4 It was Stcherbatsky's own view that the Buddhist logicians, and especially DharmakIrti, "thought that the study of lo~ic and epistemology has to replace the ancient philosophy of Buddhism." But it is not entirely clear how in Stcherbatsky's view the study of logic and epistemology was intended to lead to the goal towards which Buddhist philosophy in general is aimed. It has been said by L. W.J. van der Kuijp that the question of "the position of logic ... within Buddhist philosophy and the answer to the question whether it should playa role in Buddhism at all has been ignored by Western scholars who, taking their cue from Stcherbatsky, usually took it to be an ancilla religionis, devoid of any spiritual significance.,,6 As we have seen, this is not quite Stcherbatsky's view, nor is it a very accurate observation of modern scholarship in general, for it leaves out of account a number of scholars who have faced the question head on. One scholar it leaves out, whose assessment of the place of logic in Buddhism is not far from the one that van der Kuijp attributes to Stcherbatsky, is Edward Conze, who discussed the school of Diimaga, albeit somewhat begrudgingly, saying "both because of their historical importance, and the current interest in logic, we must briefly allude to the principles of Buddhist logic as developed by the school of Dignaga, DharmalCrrti and Dharmottara in extensive works from AD 450 (930 BE) onwards. ,,7 Conze points out that only a small minority of Buddhists took logic seriously and that it "failed to win approval elsewhere, and aroused the misgivings of many who condemned it as an utterly profane science. At variance with the spirit of Buddhism, it can indeed be tolerated only as a manifestation of 'skill in means.' Logic was studied 'in order to vanquish one's adversaries in controversy', and thereby to increase the monetary resources of the Order. ,,8 Conze apparently feels that the obscurity of Buddhist epistemologists was well-deserved, since he points out in a footnote to the passage just quoted that "these misgivings must have further increased when, observing the behaviour of people like DharmakIrti...one could not fail to notice that this branch of studies produces people who are boastful and inclined to push themselves forward." Evidently Conze does not share the view of Alex Wayman, who sees in classical Indian Buddhist debate "a dignified sportsmanship which modern debators might emulate with profit. "9 Despite his unconcealed disdain for logic and epistemology as an aspect of pure Buddhist religious practice, Conze had the highest of regard for Stcherbatsky's tome Buddhist logic, which he called "a masterpiece of

Chapter One

5

the first order, and in a class by itself," adding later that "as one would expect of a work published in Leningrad under Stalin's watchful eyes, Buddhism is here treated as a purely rational system, and the religious side ignored. ,,10 A position that Conze seems disinclined to consider is that for many Buddhists being purely rational may be the very essence of being religious. Stcherbatsky, even had he not been working under the watchful eye of a Stalinist government, may indeed have felt that Diimaga was one such Buddhist. This is a question to which we shall have to return. Conze is by no means the only counterexample to van der Kuijp's observation about modern scholars. As Ernst Steinkellner has already pointed out, the place of logic and epistemology within Buddhism was a question that T. Vetter treated at some length in 1964 in his masterful Erkenntnisprobleme bei Dharmakfrti. Steinkellner summarizes Vetter's account as follows: Valid cognitions (pramtil)ii. samyagjiiiina) are a necessary presupposition of meaningful human action. The Buddhist's actions are oriented towards the goal of emancipation. This goal and the path towards it have been shown by the Buddha. The Buddha thus offers a goal and guidance for human activity that cannot be derived from ordinary means of cognition, i.e. perception and inference. inference However, that he is an authority for this has to be proven, for faith alone is an insufficient motive to be a Buddhist. The words of the Buddha can be accepted as an authority only when it has been demonstrated that they are words of somebody who shows through his conduct that he does not lie, and who because of the development of his experience has something to tell us that cannot be mediated to us in another way. For the last goal of human actions, which also is the only point of orientation for everyday human practice, has to be indicated by such an authority, since it is never immediately present--or it would not be a "last goal". 11

In contrast to the view presented by Conze, for whom logic had only a

very questionable place within Buddhism as a rather boorish means of defeating opponents in debates for money and had a perhaps slightly more noble purpose of explaining to others what is true, Vetter and Steinkellner ~ee logic as having a place within Buddhism as a sort of "hand-maiden of theology." For them logic is in Buddhism a means that is independent of blind faith but that nevertheless serves to confirm the dogmas of Buddhism that come to be known to human beings through a process that is virtually indistinguishable from revelation. As I have argued elsewhere, this portrayal of the place of logic within Buddhism is quite accurate for DharmakIrti and later Buddhists such as S'antarak~ita and KamalasUa, but it is not necessarily accurate for Diimaga. 12

6

Preliminaries

Another modem scholar who has commented on the place of epistemology within Buddhism is A.K. Warder, who generally regards epistemology as "the essential part of philosophy" on the grounds that "other branches of philosophy are dependent on this and presuppose something known as a starting point."l Agreeing with Stcherbatsky that epistemology was regarded by Dirinaga as central to Buddhist practice, Warder spells out rather more clearly how this is thought to be so. If Asanga founded his philosophy on the Mahayana sutras and the tradition of the Tripi(aka, Dinnaga proposed to enquire first how we acquire knowledge: we cannot make any assured progress in philosophy, or, therefore, in the practice of Buddhism which is founded on knowledge of truth, unless we first ascertain what sources of knowledge can be accepted. The theory of knowledge therefore becomes for this school the basis of all study and of all practice; it replaces the old abhidharma enquiries, going to the root of all philosophical problems. 14

As can be seen from all the above quoted judgements of various modem scholars, the question of the place that epistemology and logic held in Dirinaga's overall view of Buddhism is one that remains to be answered. While it is probably impossible to muster enough evidence to convince all my colleagues on this matter, I shall nevertheless try to make a case for what strikes me after a good deal of reflection on the matter to be a likely answer to this question. And in dealing with the question of what importance Dirinaga himself gave to epistemology and logic within the context of Buddhist theory and practice, I shall also have to take into account the views of a number of Dirinaga's predecessors and contemporaries, for like any thinker Diimaga was undoubtedly very much influenced by his times and by the traditions that he inherited. Although it has generally been well-known to historians of the classical period of Indian philosophy that Dirinaga exerted a great deal of influence on the course of that philosophy, relatively little information has come down to us concerning the exact .details of Diimaga' s doctrines as he himself expounded them. This state of affairs is due largely to the fact that none of Dirinaga's more important works has survived in the original Sanskrit, and those texts that have survived at all are available only in Tibetan or Chinese translations. IS Since many of these Tibetan and Chinese translations that are now available to us show signs of having been done by translators who were themselves not certain of the meanings of many passages in the original texts, they have been unusually resistant to interpretation by modem scholars. This statement is particularly applicable to Diimaga's works in logic and epistemology. 0

Chapter One

7

The difficulty experienced by Tibetan translators as they tried to interpret Diimaga's work was probably due to several factors. First, as we can see from passages of his texts that have been recovered in their original Sanskrit, Diimaga's style is unusually terse. Second, a period of five centuries elapsed between the time that Diimaga composed his works and the time that they were first translated into Tibetan. During this period of several centuries, Diimaga' s texts were superseded by the works of other Buddhist philosophers, the most important of them being DharmakIrti, Santarak;;ita, KamalasTIa, Dharmottara, Jiianasrimitra, and RatnakIrti. Each of these subsequent thinkers expanded, modified and otherwise reformulated DiIinaga's original ideas. Consequently, by the time of their translation into Tibetan, DiIinaga's works themselves were probably rarely studied, so that translators would have had relatively few aids to help them in getting at the meanings that DiIinaga intended to convey_ Nearly all the problems that faced the Tibetan translator-scholars in their attempts to understand DiIinaga's work also face modem scholars. The problems faced by modem scholars are in some respects even more severe than those faced by their Tibetan counterparts several centuries ago. Working with any text written nearly fifteen hundred years ago is no easy matter even when there are available reliable editions of the work in the language in which it was originally composed, and the task becomes even more difficult when the original text is lost and we must try to see through the inevitable distortions of translation to what the original text might have said. Given these difficulties of reading DiIinaga at first hand, it is not at all surprising that much of what we know of DiIinaga's theories has been gleaned not from his own works but from the works of his opponents and followers. But in the case of what his opponents have reported of DiIinaga's views, it is natural to expect that it generally represents only the barest outline of his thought and tends towards oversimplification; in the case of what his followers have passed down, on the other hand, it is natural to expect that they have incorporated new developments into the theories that DiIinaga originated. Consequently, much of what has been reported of DiIinaga's ideas, coming as it does from works not of his authorship, must be regarded with caution. This cautionary note was sounded twenty years ago by Dalsukh Malvania, who, in the sole negative comment in an otherwise very positive review of D.N. Shastri'!; Critique of Indian realism, said: But it is unfortunate that the treaUnent of the Buddhist theories is not so forceful and profound as the treaUnent of the Nyaya-Vaisesika theories. The reason seems to be that the author mainly depends on the Nyaya-Vai§e~ika works for the exposition of the Buddhist theories. Only at a very few places he had used the Buddhist works, viz., K~alJa-

8

Preliminaries bhailgasiddhi, Tattvasailgraha, Samanyadil~~adilqJrasarita, Nyayabindu .... If I say that the entire work is based on the four Nyaya -Vaise~ika works, viz., Nyaya-Vartika, NyayavartikataIParyatlka, Nyayamaiijan and Nyayakandali, I am not far from the truth. 1

But while it may be a mistake when discussing Buddhist philosophy to rely too heavily on Brahmanical works to the exclusion of Buddhist works, it is no less a mistake to rely too heavily upon Buddhist works, and especially on works of the later Tibetan tradition, as guides to understanding Indian Buddhist thought.17 While all that has been said above concerning the difficulties of interpreting Diimaga is true of his writing in general, it is especially true of his most mature work, the PramtifJasamuccaya. It will be my task in this present study to recover as much as I can of Diimaga's theory of the jnterpretation of signs. Diimaga discusses the interpretation of inferential signs (Zinga) in general in the second chapter of PramtifJasamuccaya, while in the fifth chapter of that work he deals with articulate speech as a special kind of inferential sign. It is in this fifth chapter that he expounds the doctrine of linguistic meaning known as apohavtida, a theory of language that excited interest among Indian philosophers for over half a millennium after it was first introduced. The aim of the present work is to help clarify this doctrine as it was known just to DiIinaga, quite independently of his later interpreters, and to examine the place of Dirinaga's theories of linguistic signs within his overall philosophical system. The main sources on which this study is based are: a) the two extant Tibetan translations of the PramtifJasamuccaya" b) the Tibetan translation of Jinendrabuddhi's extensive commentary on this work, and c) the works of several a.uthors who wrote at a time near to DiIinaga's own and who quoted or extensively paraphrased his views. Most important in this last group are the MImarilsa philosopher Kumarila Bhana, the Jaina philosopher Mallavadin K~amasramaI).a, and the Naiyayika thinker Bharadvaja alias Uddyotakara. While no one of the above sources by itself would suffice to give us a complete and accurate picture of Dirinaga's presentation of apohavtida, it is hoped that by using them together we can gain a better understanding of his doctrine than has been generally available up to the present.

Chapter One

9

1.2 The contributions of previous scholars A study sUGh as this one becomes feasible only wheI! a great deal of groundwork has already been done. And so in order to reveal 'some of the foundation upon which the present study is built, my study begins with a survey of the work done from the time that Diimaga was rediscovered by modern scholars until the very recent past. In this introduction I shall discuss in any detail only those works that have added significantly to our understanding of Dmnaga's doctrines concerning inference and language and shall mention only briefly, if at all, the many later works that have drawn upon these sources. Not surprisingly, not all of what the pioneering scholars said about Diimaga on the basis of their preliminary researches is supported by a more detailed look at DiIinaga's own work. One of the principal functions of this survey of earlier scholarship must therefore be to sort out some of the specific interpretations of his philosophy that have become outdated and in so doing to indicate some of the misconceptions about Diimaga's system of philosophy that pioneering scholars have inadvertently passed down to us. By disencumbering ourselves of these misconceptions at the outset, I hope to render our task of exploring DiIinaga's philosophy with a fresh mind considerably less difficult. Having argued that many of the earlier scholars have given us a rather inaccurate and incomplete view of DiIinaga's theories of inference and language, owing primarily to their over-reliance on second-hand accounts of his thought, my task in subsequent chapters of this work will be to try to give a more comprehensive account of DiIinaga's system of logic, epistemology and ontology as presented mostly in his Pramli1)asamuccaya but also in some of his other works.

1.2.1 Satis Chandra Vidyabhii~aJ)a J1le study of the schools of logic and epistemology developed by the Buddhists of classical India is, relative to the study of other aspects of Buddhist philosophy and indeed of Indian philosophy as a whole, a rather recent enterprise among scholars writing in modern European languages. The modern study of Indian Buddhist epistemological theory did not really begin until the beginning of the twentieth century, when Satis Chandra Vidyabhii~aI}.a published several articles announcing the findings of his preliminary examination of the section devoted to .logical works in the Tibetan Bstan 'gyur, copies of which had only recently been taken to Calcutta by the Tibet Mission. These reports appeared in the years 1905-1907.

10

Preliminaries

It was Vidyabhu1)al).a who first recognized the importance of this collection of logical treatises in the Bstan ' gyur, all of which had originally been composed in Sanskrit and later translated into Tibetan. He correctly saw them as the key in the task of piecing together the history of Indian logic in the period between the time of the ancient Nyaya commentators and the time of GailgeSa, the thirteenth century thinker traditionally regarded as the founder of modern Nyaya. Indeed Vidyabhii~aI).a argues that modern Nyaya, which is characterized by the study of purely logical and epistemological topics in contrast to the ancient Nyaya that discusses not only logic but also metaphysics, theology and soteriology, really had its beginnings in the sixth century with Diilnaga rather than in the time of Gailgesa. Vidyabhu~al).a writes: Having considered the sixteen categories treated in the ancient Nyaya to be redundant and some of them as mainly based on the orthodox principles of the Hindus, the Buddhists took up only one category, viz., Pramaf}a (evidence of knowledge), and treated it in such a way that the doctrine of evidence might be equally applied to the religious systems of the Hindus and the Buddhists. The attempt on the part of the Buddhists to divest the principle of logic from those of theology, metaphysics, etc., was the cause of the foundation of the modem Nyaya, otherwise called Tarkasastra or Logic proper. 18

In this same article Vidyabhii~al).a offers an account of Diilnaga's career, uncritically following the rather fanciful biography handed down by Taranatha. More importantly, he correctly identifies the name Chenna, which occurs in the accounts of Yijing and other Chinese travellers to India, as the Chinese rendering of Diilnaga's name. 19 Still more important, Vidyabhu~a1}a cites several more sources, including Uddyotakara and Vacaspatimisra from the Nyaya camp, and Kumarlla Bhatta and Parthasarathimisra from the MImamsa school, where Diilnaga's views are summarized and subjected to criticism. Moreover, he draws attention to several Buddhist authors who expanded Diilnaga's basic views, notably Dharrnaklrti, Dharrnottara, VinItadeva, Santabhadra, Alailkadeva, Jinendrabuddhi (or Iinendrabodhi) and KamalasTIa. Further comparisons of the Tibetan translations of Diilnaga's PramalJasamuccaya with the extant Sanskrit texts of Diilnaga's rivals later led Vidyabhii~aI)a to the discovery of several passages of Diilnaga's text that had been quoted by later Sanskrit philosophers. 2o This process of discovering the original Sanskrit wording of passages in Diilnaga's works and matching these fragments up with the extant Tibetan passages has made considerable progress since Vidyabhii~a1}a's time through the efforts of such scholars as Randle (1926), H.R. Rangaswami Iyengar (1927), Tucci (1928), Frauwallner (1929), Durgacharan Chatterji (1929) and Katsura (1975), but it is largely owing to VidyabhU~aI).a's preliminary investigations that subsequent

Chapter One

11

scholars have known where to begin their efforts to carry out more thorough searches for fragmentary quotations of Diimaga's original words. Vidyabhii~al).a offered little more than a sketch of Dirinaga's theory of inference and theory of languageY But as sketchy as these accounts were, they provided an important impetus to future generations of scholars to work on the relevant texts more closely. Despite this good start in the study of Indian Buddhist logic and epistemology, a full quarter of a century passed from the time of Vidyabhii~al).a's first bringing Dirinaga to the attention of modern scholars until the appearance of the first detailed study of the system of philosophy that is generally thought to have been founded by Diri.naga. That study was achieved by the Russian scholar Stcherbatsky, to whose work we shall now turn.

1.2.2 Th. Stcherbatsky It is probably still true that the author whose works are most widely known in the area of Indian Buddhist logic and epistemology is Th. Stcherbatsky. His major contribution to the field was his two volume study Buddhist logic, which was based primarily on DharmakIrti's Nyayabindu, a relatively minor work but one that was well suited to serve as an introduction to the issues and techniques of Buddhist logic and epistemology. In elucidating some of the key ideas expressed in that work, Stcherbatsky attempted to place them in their proper historical context by making frequent references to Dirinaga's system of thought as well as to some of the discussions of post-Dharmakirtian Buddhist and Brahmanical philosophers. The second volume of Buddhist logic contains a complete English translation of DharmakIrti's Nyayabindu along with Dharmottara's commentary, and translations of excerpts from works by the Buddhist philosophers Vasubandhu, VinItadeva, Dirinaga and Jinendrabuddhi as well as from the works of the Brahmanical thinkers Vacaspatimisra and Udayana. Thus we have in Stcherbatsky a reasonably thorough outline of the development of Buddhist logic from, the early sixth century, which was Dirinaga's period of activity, up to the late eighth century, which was Dharmottara's period. In addition we have an account of Brahmanical reactions to Buddhist thought up to the time of the tenth century scholar Vacaspatimisra. Given the importance of Stcherbatsky's seminal study of the Indian Buddhist epistemologists and their place within Buddhist history as a whole, it is worth examining his views on the subject in some detail.

12

Preliminaries

It is characteristic of Stcherbatsky's wntmgs on the history of Indian philosophy that he depicts nearly all intellectual movements as deliberate and systematic rejections of trends within immediately preceding intellectual movements. Thus it is not surprising to find that he sees the origins of Buddhist logic as a systematic reaction to two very different earlier tendencies within Indian philosophy, namely, the "extreme skepticism" and mysticism of the Madhyamakas and the "naive realism" of the ancient Nyaya and Vaise~ika systems. We see, for example, that Stcherbatsky characterizes the early Mahayana, of which the Madhyamaka is his prime example, as holding the view that "the only source of true knowledge is the mystic intuition of the Saint and the revelation of the new Buddhist Scriptures, in which the monistic view of the universe is the unique subject. ,,22 Still on the subject of the Madhyamakas he says a few lines later: "This is a further outstanding feature of the new Buddhism, its merciless condemnation of all logic, and the predominance given to mysticism and revelation." Further on Stcherbatsky says that the later Mahayana, on the other hand, represented by Asailga, Vasubandhu and their more "moderate" followers, Dirinaga and DharmakIrti, were intensely interested "in logical matters, and that this interest in logic "becomes overwhelming and supersedes all the former theoretical part of Buddhism." Moreover, the interest in logic seems "to have been directed against the extreme skepticism of the Madhyamakas. "23 But it was not only in opposition to these anti-logical Madhyamakas that Dirinaga formulated his doctrines, says Stcherbatsky, as in his discussion of the Nyaya- Vaise~ika system he tells us: "Buddhist logic was created in a spirit of decisive opposition to the logic of these Realists. ,,24 And he sums up his remarks on Nyaya- Vaise~ika in these words: The school of Nyaya had already developed logic when the Buddhists began to manifest a keen interest in logical problems. The Buddhist doctrine then came to graft itself on the early pre-Buddhist stock. But then a clash supervened at once between the two utterly incompatible outlooks. The brahmanical logic was formal and built up on a foundation of naive realism. The Buddhists at that time became critical idealists and their interest in logic was not formal, but philosophic, i.e., epistemological. A reform of logic became indispensable. It was achieved by Dignaga. 25

Thus, it is clearly Stcherbatsky's view of the historical development of the ideas that we find expressed in the works of Dirinaga and his followers that the Buddhist logicians, reacting against the MMhyamaka tendency to regard all discursive and analytic thought as unsuited to the task of leading one to higher wisdom and therefore unworthy of serious pursuit. sought to re-establish reliable criteria for distinguishing sound argumentation from unsound, and true cognitions from error. But in reaction to the Nyaya and

Chapter One

13

V aise~ika insistence on accepting such things as universals, relations and the solil as ultimately real, the Buddhist logicians sought to establish a more austere ontology, relegating universals and relations to the realm of theoretical constructs that, while of undeniable utility in carrying out analytical thought, were lacking ultimate reality. The doctrine ofapoha, which Stcherbatsky calls "Nominalism," emerges as part of this campaign for metaphysical parsimqny. Before going any further, let me examine the extant to which Stchetbatsky's general picture of the place of Buddhist logic in Indian philosophy is an accurate picture of Diimaga's place in Indian philosophy. First, how accurate is it to say of Dirinaga that his philosophy is a systematic reaction to Madhyamaka skepticism? This is a rather difficult matter to assess, for we find in his Pramtiljasamuccaya no explicit references to Madhyamaka notions, and we certainly find no systematic attempt to come to terms with arguments delivered by Nagarjuna or subsequent MMhyamaka authors. Thus if Dirinaga was indeed motivated by a desire to counter the extreme skepticism of those Buddhists who preceded him, he at least did not ostensibly build his text upon the framework of that motivation. Not only is there no evidence in what DiiJ.naga says explicitly for his having taken an anti-Madhyamaka stance, but, as I shall argue in the subsequent chapters, a case can be made for seeing Dirinaga as a natural development of trends that were part of early canonical Buddhism as well as of the Madhyamaka movement; it is, in short, not necessary to see Dirinaga's epistemology as a reaction to anything within the Buddhist movements that had preceded him. Second, on the matter of Dirinaga's disputes with the Naiyayikas and Vaise$ikas, an examination of his critiques in Pramtiljasamuccaya of their views shows that he confines himself primarily to two formal issues. First, he assesses whether or not their fundamental definitions are adequate to define what they purport to define and whether or not the set of basic defmitions advanced by a rival system are logically consistent. And second, he attempts to anticipate and answer whatever objections these rival systems might raise concerning the adequacy and consistency of his own basic definitions. We find on Dirinaga's part remarkably little participation in the conflict that Stcherbatsky reports between the so-called "Idealists" and "Realists," although there is no doubt that many of Dirinaga's views helped to precipitate the conflict among the Indian philosophers of subsequent generations. Now there is one more point to make with respect to Dirinaga's place in Indian philosophy, and that is on the matter of the influence of the Vedantin grammarian Bhartrhari's theories on Dirinaga's views on

14

Preliminaries

language. About this more will be said later (see 1.2.6); for tl:1e present all we need say is that it is a very important factor in DiIiniiga's thought of which Stcherbatsky was apparently unaware and that this unawareness tends to distort Stcherbatsky'saccount of DiIiniiga's apoha theory of meaning. Let us now tum to Stcherbatsky's interpretation of that theory. Unfortunately, his account of the role of apoha within DiIiniiga's system is rather opaque, and on several points where it is clear it happens not to be a very accurate account of DiIiniiga's actual theory. Stcherbatsky's inaccuracy consists in essentially two things, namely, a) his attributing'to DiIiniiga doctrines that were in fact advanced by his successor DharmakIrti, and b) his misconstrual of particular passages of DiIiniiga's own work. In order to illustrate the fIrst of these two forms of inaccuracy, I must fIrst offer a succinct account of how Stcherbatsky sees DiIiniiga's overall system. At the outset, says Stcherbatsky, the Buddhists draw a radical distinction between two realms of cognizable objects. On the one hand there is the realm of particulars characterized by a series of absolutely distinct "point-instants," which are unintelligible and inexpressible.26 They enter into our awareness as the contents of sensation. And on the other hand there is the realm of universals characterized as mental constructions or conceptsconcepts. 27 These alone are expressible, and they enter into our awareness solely as the contents of judgement and inference. The essential nature of a concept is "negative" 'or "dialectical" insofar as it functions to differentiate the particular that caused it from all those particulars that could not have caused it. Of these two realms, the realm of particulars and the realm of universals, only the realm of particulars may be regarded as ultimately real, the criterion of reality being causal efficiency, which practically reduces to the capacity to satisfy material needs. 28 Now, given that the content of any judgement or inference is a universal that lacks this causal efficiency, it follows that in one ~ense all judgements are erroneous, for they fail to have reality as their subject matter. There is, nevertheless, a test by means of which one can distinguish accurate judgements from inaccurate ones, and that is the test of practicality. For the primary role of concepts is to guide our actions in the world of particulars. Thus if a given concept guides an action that turns out to have an expected consequence in the form of a material desire being satisfied, then that concept may be called accurate or correct. But if it guides an action that turns out to have an unexpected consequence, it may be deemed inaccurate or incorrect. Hence the ultimate test for the accuracy of any inference is whether or not the idea produced by that inference leads to successful action. Essentially

Chapter One

15

the same pragmatic test holds for determining the accuracy or truth of linguistic expressions. For language is just a means by which a speaker communicates one of his judgements. A speaker makes a statement that results in a hearer's forming a judgement; if the hearer acts on that judgement and obtains expected results, the statement can be said to have been reliable or true. Otherwise, not. Thus a statement in language has as its content a concept that has essentially the same function as a concept in a privately held judgement. As we saw above, that nature is essentially "negative" or "dialectical." I have said that Stcherbatsky's account of Diimaga's system, the essential points of which I have tried to outline above, imputes to Diimaga views that he did not explicitly assert. The most important of these are: 1. the view of particulars as point-instants, which amounts to a

commitment to a doctrine of radical momentariness (ksat;ikavada), 2. the notion of causal efficiency as the criterion of a real thing, which, alongside the view that universals lack causal efficiency, results in the assertion that all cognitions having universals as their contents are intrinsically erroneous, and 3. the pragmatic test for the accuracy of judgements, inferences and statements. These doctrines, along with other doctrines concerning such questions as how and why concepts are formed, were the innovations of DharmakIrti, who developed them is response to criticism of Diimaga's theories. But Diimaga himself, as we shall see below, was primarily concerned with an altogether different set of issues. Stcherbatsky's failure to distinguish properly the system of Diilnaga from that of his followers--a failure that is quite understandable considering the state of knowledge at his time--has set an unfortunate trend in the modern study of classical Buddhist philosophy, that being the tendency to assume that DharmakIrti and subsequent Buddhist logicians were faithful representatives of the views of Diilnaga. In varying degrees this tendency has been followed by S. MookeIjee (1935), D.N. Shastri (1964), D. Sharma (1969), RR Dravid (1972) and C.L. Tripathi (1975). The 1975 article by Tripathi is a particularly outstanding example of a failure to distinguish properly between Diilriaga's thought and DharmakIrti's. It is entitled "The role of apoha in Dignaga's theory of knowledge," but despite the title's focus on Diilnaga, Tripathi cites Diilnaga only once,

16

Preliminaries

and that only indirectly, since the citation is taken from Stcherbatsky's Buddhist logic. All other textual citations in Tripathi's work are to paraphrases of Diimaga's views by Kumiirila Bhana, Uddyotakara, Santarak~ita and to Buddhists and members of the Nyaya school who even come after Siintarak~ita. Needless to say, Stcherbatsky himself cannot be held responsible for all the times his failure to separate Diimaga from DharmakIrti and his various other errors have been repeated by others. Given, however, that Buddhist logic remains at present the standard work on the philosophy of the school founded by Diimaga, and given that so many scholars have repe~ted Stcherbatsky's misconceptions along with his more sound observations, it is perhaps necessary to say that his work is in serious need of re-examination and will, it is hoped, soon be replaced by a more accurate study based on all the advances that have been made in the study of the Buddhist epistemologists since 1932.

1.2.3 Satkari Mookerjee At the same time that Stcherbatsky and Frauwallner (see 1.2.4 below) were working in Europe on various aspects of the philosophical system of the Buddhist epistemologists, Satkari Mookerjee of Calcutta University was also working on a book, The Buddhist philosophy of universal flUX, that, like Stcherbatsky's Buddhist logic, treated the philosophical system of the Buddhist epistemologists as a whole and sought to show the interrelations of the various individual doctrines propounded within that system as well as the relations of that system with other Buddhist and Brahmanical systems of thought. The subtitle of Mookerjee's book is "An examination of the philosophy of critical realism as expounded by the school of Dignaga." In his preface Mookerjee claims that his work "is an humble attempt to give a critical exposition of the Medieval School of Buddhism that was ushered into existence by Dignaga and DharmakIrti and later on systematized by Santarak~ita, KamalasUa, RatnakIrti and other authors of repute. ,,29 On the next page he says his work is "chiefly preoccupied with the dry metaphysical and epistemological sides of the Sautrantika philosophy." It is this Sautriintika philosophy that Mookerjee characterizes as the school of "critical realism" in contrast to the "naive, common-sense realism" of the Nyaya-Vaise~ika and MImiiIhsa schools, the "self-complacent realism" of the Sarvastivlidins, the "subjective idealism" of the Yogaciira school and the "absolute negativism" of the Madhyamaka schoo1.3o All of this seems to imply that Mookerjee regarded Diimaga and his followers to be adherents of the Sautriintika school of Buddhism. Since, as we shall see below, discussions of sectarian rivalries are a prominent feature of Mookerjee's inteipretation

Chapter One

17

of doctrines presented by Buddhist epistemologists, it is important for him to make clear his position on Dinnaga's sectarian affiliations and to defend that position as well as the textual evidence will allow. This approach that Mookerjee takes is, I believe, of little value in dealing with logical and epistemological issues, for the simple reason that the doctrinal issues that differentiated the schools of Buddhism are not the issues under discussion in the works of the Buddhist epistemologists, and especially in the work of Dinnaga. 31 The epistemological treatises reveal little about their authors' beliefs in matters other than epistemology, and it is not even very clear that knowing how these authors stood on most controversies within Buddhism would reveal much if anything about their epistemological views. This is especially true of Diimaga's PramalJasamuccaya, a work that is remarkably free of sectarian bias. For this reason, although it seems perfectly justifiable to deal as Mookerjee has done with Dmnaga, Dharmaki:rti, Dharmottara and others as a group of thinkers who had a common set of philosophical interests, I see every reason to be cautious in identifying the Buddhist epistemologists, and especially Dinnaga, with the Sautrantikas or any other sect within the Buddhist community. Above I said that Mookerjee's approach to studying the Buddhist epistemologists was like Stcherbatsky's in that he tends to treat the school as a whole as ifit were a unified movement. Also like Stcherbatsky, but to a much greater extent than the Russian scholar had done, Mookerjee focused his attention on the polemics exchanged between the Buddhist thinkers and their philosophical opponents, chiefly the Naiyayikas and the MImamsakas, but also the Jainas, the Sfuhkhyas and the Vedantins. According to Mookerjee, polemics almost constitutes the life-blood of Indian philosophy. He says, for example, "In fact, [the Indian philosophers] fought for what they believed to be a question of life and death. Philosophy was not a matter of academic interest in India. ,,32 In keeping with this view of the paramount importance of sectarian rivalry in medieval Indian philosophy, Mookerjee depicts the dispute between the rival systems in dramatically bellicose· terms; they were "daring and desperate fights ... ,which were fought on a hundred and one battlefields." In the same paragraph he describes Kumarila Bhatta as "a hero of a thousand and one battlefields," Ita veritable Tartar" who delivered "sledge-hammer blows" against his Buddhist opponents. 33 Such a view of Indian philosophy places, I believe, too much emphasis on sectarian rivalries and too little emphasis on the intellectual curiosities of individual thinkers. My own preference would be more to concentrate on the struggles of individual thinkers--not their struggles against philosophical "enemies,"

18

Preliminaries

but rather their struggles with some very tough philosophical problems that plague anyone who thinks about them, regardless of one's philosophical or religious persuasions. Seen in this light, one philosopher's criticisms of another's views is not merely a de rigueur exercise in team spirit and loyalty to his school, but rather a more sincere attempt to understand how one professed solution to a philosophical problem may have intrinsic shortcomings that, unless improvements are made to overcome them, may disqualify it as the best solution to the problem at hand. That MookeIjee elected to depict the history of Indian epistemology more as a conflict of Buddhist against Brahmanist is not surprising when one considers that his most often cited source of information on the Buddhist thinkers is the encyclopedic Tattvasangraha of Santarak~ita with its commentary by KamalaSIla, a work that attempts to reconcile all the disparate Buddhist views and defend them collectively against the criticisms of non-Buddhist philosophers. Mookerjee's work is divided into two main parts, of which the first deals primarily with metaphysical issues and the second part with logic and epistemology. It is in the flrst of these two main parts, in which Mookerjee discusses various ontological issues as well as various theories of the soul and various conceptions of nirvar;a, that he treats the apoha theory of meaning. For MookeIjee it is clear that the apoha doctrine arises out of the Buddhists' unrelenting denial of anything permanent in the universe, which in tum prompts a denial of universals as existing independent of the mind that conceives them. The gist of the Buddhists' argument against the extra-conceptual reality of universals is put by Mookerjee as follows: Now we deny the existence of universals because there is no proof in their favour. Universals are posited to account for conceptual thoughts. But no causal relation can be discovered between concepts and universals. Causal relation is understood by means of the Joint Method of Agreement and Difference. But universals being eternal verities and conceptual thoughts being occasional events, there can be no causal a particular concept can relation between them. The non-emergence not be due to the absence of the universal concerned, as universals without exception are present always. Nor can the occasional emergence of a conceptual thought be causally affIliated to a universal, because the universal is ever present and if it had such efficiency, it would generate the idea always. So nothing is gained by postulating universals. If, however, the cognition of universals is supposed to be contingent on the cognition of the particulars in question, we do not see what these effete universals will avail. Our conceptual thoughts are seen to arise even without them.34

or

19

Chapter One Thus for Mookerjee's Buddhist constructions pure and simple. "35

"universals

are

but

subjective

Having discussed the Buddhist position on the ontological status of universals, Mookerjee next deals with the question of what it is that a word "connotes," a question that loses some of its clarity in Mookerjee's discussion since he uses the terms "connotes," "refers to" and "denotes" interchangeably with no apparent regard to the technical uses of these terms by most writers on semantics. In this connection he summarizes several Buddhist arguments against various candidates that present themselves as the possible meaning of a word. First the Buddhist rules out particulars as the meanings of words or symbols on the grounds that the usage of symbols requires an established convention within the community of users of those symbols, and, since particulars are momentary, a particular would vanish before any convention linking it to a symbol could be established. 36 But if a word does not have a particular as its meaning, then, since particulars are the sole realities accepted by the Buddhists, it follows that for the Buddhist "words have no reference to reality in any sense. Words in [the Buddhist's] opinion deal with concepts and these concepts are purely subjective constructions. ,,37 I have, incidentally, considerably condensed the arguments outlined by Mookerjee, but I think the essential features of the arguments have been preserved in the simplification. Mookerjee also deals with the Buddhist contention that all relations are fictitious owing to their having no existence apart from the relata to which they pertain. Similarly, he represents the Buddhist as saying that the notions of an individual (vyakti) and configuration (akrti) are "conceptual vagaries" and "unsubstantial fictions, pure and simple." Having demonstrated that the Buddhists hold that words do not refer to any external reality, Mookerjee represents them as holding the position that "what is signified by a word is neither a subjective idea nor an objective reality, but something fictitious and unreal, which is neither here nor there. The fact of the matter is that both the speaker and the hearer apprehend in fact and reality a mental image, a subjective content and not any objective fact. ,,38 Mookerjee then informs us that each mental image is distinct from every other, and for this reason "we characterize [the mental image] by a negative expression, viz., 'negation of another' (anyapoha).,,39 Hence for the Buddhist "the connotation of a word is rightly looked upon and designated as a fundamental negation." 40 Having offered this account of the essential ingredients in the basic apoha theory of meaning, MookeIjee goes on to show how it was criticized by Uddyottakara, Kumar-ila Bhatta,

20

Preliminaries

and Bhamaha, reformulated by Santarak~ita, criticized again by Vacaspatimisra and Jayanta Bhatta, and reformulated a second time by Ratnaklrti. It should be clear from the foregoing truncated account of Mookerjee's presentation of the doctrine of apoha that he regards the principal impetus of the doctrine to be metaphysical; the Buddhists, having denied the reality of universals, are compelled to account for all those facts of everyday experience that realists account for by an appeal to the existence of universals, among those facts of everyday experience being the workability of language and inference. Incidentally, this emphasis on metaphysical doctrines as the basis of the overall system of the Buddhist epistemologists is followed up in D.N. Shastri's Critique of Indian realism and R.R. Dravid's Problem of universals in Indian philosophy. The latter says: It is clear... that the whole controversy between the realist and the Buddhist nominalist is a clash between two radically opposed metaphysical standpoints. All their disputes over logical and epistemological issues are traceable to their respective metaphysical presuppositions. The metaphysical assumption underlying the realist's position is the conception of the real as determinately knowable and expressible.... The assumption underlying the Buddhist position, on the other hand, is the conception of the real as the unique point-instant of causal efficiency.41

Mookerjee's work, like Stcherbatsky's, provides a good overview of the system of the Buddhist epistemologists who came after Dinnaga, and being rich in quotations from authors who wrote after DharmakIrti, it is also a good beginning on the sort of project that Frauwallner (see 1.2.4) had hoped to undertake, namely a history of the development of ideas within the movement of Buddhist logicians. But, again like Stcherbatsky, Mookerjee tends to treat the doctrines of the early phase of the school as a whole, with the result that the reader is left with the mistaken impression that Dinnaga and Dharmaldrti advanced essentially the same lines of reasoning.

In the above discussion on Stcherbatsky we have already reviewed several metaphysical doctrines that are central to DharmakIrti's arguments concerning the apoha theory of meaning but that Dinnaga does not mention at all. To what was said there (p. 15) we can, in the light of Mookerjee's recounting of the arguments of "the Buddhists," add some further points of difference between Dinnaga and DharmakIrti.

Chapter One

21

1. The line of reasoning cited above (p. 18) concerning the Buddhist denial of the reality of universals is not to be found in DiIinaga's work. . 2. The line of reasoning cited by Mookerjee concerning the denial of particulars as the meanings of words is one that DiIinaga himself does not employ. DiIinaga does deny the reality of universals, but his reasons for doing so are very different from the ones cited by MookeIjee. Similarly, DiIinaga argues that a particular cannot be the meaning of a word, but again he advances an entirely different set of reasons for that position from those reported by Mookerjee. Since DiIinaga's arguments will be discussed in detail later in this work, we shall note here only that they differ from those of his successors and hence from those cited by MookeIjee. Thus we can say of Mookerjee, as we said of Stcherbatsky, that his work turns out not to be particularly accurate as a source of information on DiIinaga's own manner of treating the philosophical issues that were the general concern of the Buddhist epistemologists. A clearer and more accurate picture of DiIinaga's philosophy as expressed by DiIinaga himself was not, however, to appear until nearly a quarter of a century after MookeIjee's Buddhist philosophy of universal flux. 1.2.4 Erich Frauwallner

Stcherbatsky's presentation of the most important features of the philosophy of the Buddhist epistemologists had been based, as we saw in section 1.2.2, on a relatively minor work of DiIinaga's earliest important interpreter, DharmakIrti. By studying the Nyayabindu along with portions of works by later authors, Stcherbatsky was able to offer an overview of the philosophical system of the Buddhist epistemologists in general, and particularly of Dharmaldrti's system. For a more detailed understanding of DharmakIrti's philosophy, however, it is necessary to tackle his most important work, Pram(1)avarttika, wherein a number of issues either not treated in the Nyayabindu at all or dealt with there only superficially are treated in considerable detail. One of the earliest scholars to engage himself in a study of the Prama1)avarttika was the Austrian scholar Erich Frauwallner, who in 1930 began publishing a series of articles on the apoha theory of meaning as presented in the Pram(1)avarttika and in an independent work by Dharmottara, the Apohaprakara1)am. 42

22

Preliminaries

When Frauwallner fIrst worked on the Pramal)avarttika, the text was believed to be extant only in Tibetan translation, the Sanskrit original not having yet been discovered by any modern scholar. Thus in the first of his articles on the apoha theory, Frauwallner presented an edition of the Tibetan translation of verses 42 through 187 of the Svarthiinumanapariccheda of the Pramal;lavarttika. 43 For a little less than one quarter of this set of verses he was also able to cite the original Sanskrit as quoted in Sakyabuddhi's subcommentary. Next, in 1932 and 1933, Frauwallner published a German translation of these verses along with a summary of each verse based on a study of the Sanskrit commentaries to the verses translated. In 1935 Frauwallner concluded this phase of his work on DharmakIrti's apoha theory with an article summarizing the main philosophical issues and lines of reasoning contained in these verses and tying the apoha theory into DharmakIrti's overall system. The year 1930, in which Frauwallner began this work, was, incidentally, the same year that Stcherbatsky published his translation of the Nyayabindu, and the year that Frauwallner published his overview of DharmakIrti's apoha theory was the same year that Satkari Mookerjee published his Buddhist philosophy of universal flUX. The early 1930's were therefore very important years for the renewed study of DharmakIrti's logic and epistemology. Frauwallner's work on DharmakIrti was followed in 1937 by an edition of the Tibetan translation of Dharmottara's monograph on the apoha theory, the Apohaprakaral;la (Gzhan set ba rab tu byed pay, along with Frauwallner's German translation and comments. Whereas the tendency in Stcherbatsky's presentation had been to treat the doctrines of the Diimaga school as a whole, presenting some three hundred years of philosophical development almost as a unified body of thought, Frauwallner seems to have been more interested in discussing the achievements of individual thinkers. It had been his intention, in fact, to produce a history of the evolution of the apoha theory from Dirmaga's time to at least Dharmottara's time, with the work on Dharmaldrti's contribution to that theory being a frrsti.nstalment in this larger project. In his introductory remarks to the 1930 article, Frauwallner points out that while the apoha theory originated with Diimaga, DharmakIrti's passage on apoha is not merely an elucidation of earlier ideas but rather a new creation embodying the results of Dharmaldrti's own thinking. In this connection Frauwallner realized quite early that DharmakIrti's doctrines, while based on ideas presented in Diimaga's work, represented the beginning of a considerable evolution and modification of those ideas, an evolution made necessary in order to counter the many cogent criticisms of Dirinaga's original ideas by Kumarila Bhatta and Uddyotakara, then later to counter the criticisms of DharmakIrti' s ideas by later Brahmanical

Chapter One

23

thinkers. 44 It is in this evolutionary framework that Frauwallner discusses such features of Dh!l1lllaldrti's thought as his commitment to the doctrine of radical momentariness of all particulars,45 the notion of causal efficiency as the criterion of a real thing,46 the pragmatic test for the accuracy of judgements, and the psychological question of how general concepts are formed. 47 Although Frauwallner managed as early as 1935 to separate DhannakIrti's achievements from those of his predecessor Diimaga, it was not until 1959 that he was able to turn his attention to an examination of all of Dirinaga's extant works with the aim of establishing a relative chronology of those works and thereby charting the evolution of Dirinaga' s thought. The result of this work was his article "Dignaga, sein Werk und seine Entwicklung." By noting which of Dirinaga's .works were cited by name in other of his works, and then by comparing passages that recurred either verbatim or with slight modifications in several of his works, Frauwallner was able to trace the revisions and refinements that Dirinaga made in his treatment of logical and epistemological matters and to show that the Pramii1]asamuccaya represents the culmination of Dirinaga' s work in the field of logic and epistemology. And he concluded further that this interest in logic and epistemology represented the final phase of a long career that had begun with essentially apologetic studies of first the Yogacara school of Buddhism and then the Sautrantika school. While working in Sautrantika abhidharma, suggests Frauwallner, Dirinaga began to develop a number of his ideas in epistemology. Also in this middle part of his career, concludes Frauwallner, Dirinaga undertook critical studies of the Sarhkhya, Vaise~ika and Nyaya of orthodox Brahmanism. 48 Frauwallner was also able to show that one of Dirinaga's earliest works, the Traikiilyaparfksii, was modelled very closely on a work by the Vedantin grammarian Bhartrhari, in whose work we can also detect defmite foreshadowings of Dirinaga's apoha theory.49 Frauwallner's work, then, constituted a significant advance in the study of Dirinaga as an individual philosopher and has provided the most useful framework to date for the study of Dirinaga' s individual works and doctrines. . In the course of his study, Frauwallner was able to shed light on the apoha doctrine, a doctrine that he says is foreshadowed in one of Dirinaga's early works on proof and refutation, the Nyiiyamukha, and may have been worked out in greater detail in works now lost to us, but which is given its fullest treatment in the fifth chapter of the Pramii1)asaml,lccaya. 50 In reading through this fifth chapter, Frauwallner discovered a number of striking parallels between the chapter on apoha and the second chapter of the Pramii1)asamuccaya, which deals with inference for

24

Preliminaries

oneself (svartMnumana) , and concluded therefore that the 'doctrine of apoha grew out of Diimaga's theory of inference. 51 What was special about Diimaga's theory of inference was, according to Frauwallner, that for Diimaga, in contrast to earlier theories of inference such as we find in the Nyaya and Vaise1;>ika systems, inference was based not upon an appeal to causal relations between the inferential sign (lifzga) and the thing inferred from it, but rather upon the relative scopes of concepts, whereby wider concepts could be inferred from narrower but not vice versa. 52 If a narrower concept is included within a wider concept, then the presence of any particular that is within the scope of the narrower excludes the applicability to that particular of any concept that is contradictory to the wider; the essential nature of the information that is produced by an act of inference is the general fact of the exclusion (apoha) or inapplicability of some concepts given the applicability of certain other concepts. Even Diimaga's definition of sensation (pratyak$a) as a cognitive act that is free from lhe application of concepts to particulars was derived, in Frauwallner's view, from this realization that there are two mutually distinct types of cognitive objects, namely, particulars and concepts. This led Diimaga to assert that there are two and only two kinds of cognitive act, namely, one that has particulars as its subject matter and one that has the application of concepts as its subject matter. 53 Thus we have in Frauwallner a very different notion of the role of the apoha doctrine in Diimaga's system from that which we found in Mookerjee. For MookeIjee the apoha doctrine was necessitated by a denial of the objective reality of universals; in this view, metaphysical questions are of primary importance in the sense that once a certain metaphysical stand is taken, it places certain constraints on the type of logical and epistemological doctrines that can consistently advanced. For Frauwallner, on the other hand, the thing of primary importance to DiIinaga was his understanding of inference, from which a denial of the objective reality of universals emerged as an almost incidental by-product. About these two different views of the relation of the apoha doctrine to metaphysical doctrines more will be said later. 1.2.5 Kitagawa Hidenori, Hattori Masaaki and Katsura Shoryu It has already been noted that a number of scholars have failed to distinguish between DiIinaga's philosophical views and those of his interpreter DharmakIrti, despite the fact that Frauwallner had pointed out as early as 1935 that DharmakIrti had made a number of significant

Chapter One

25

deviations from DiIinaga's original theories. The failure to distinguish the two thinkers is understandable, however, when it is recalled that DiIinaga's works on epistemology do not survive in the original Sanskrit and are extremely difficult to read in their Chinese and Tibetan translations. Credit for making important advances in the enormously complex task of deciphering DiIinaga's philosophy as DlIinaga himself expressed it must go to Kitagawa Hidenori,54 who provided Japanese translations of parts of the key chapters of DiIinaga's most important work, the Pramaijasamuccaya, a treatise that is particularly difficult to recover because of the fact that the two Tibetan translations are often quite different from one another. Before any progress could be made in piecing together the puzzle of the Pramaijasamuccaya it was necessary to make decent editions of the two Tibetan translations of this text and to correlate Tibetan passages in these texts with original Sanskrit versions as preserved in quotations in extant Sanskrit works. Kitagawa did not provide a true critical edition of the Tibetan translations of Pramaijasamuccaya, but he did provide what he called a "faithful copy" of key passages from the translation by Vasudhararak~ita that appears in the Sde dge Bstan 'gyur, which he copied together with the translation by Kanakavarman that appears in the Peking edition of the Tibetan Tripitaka the two versions were placed side by side in double columns for ease of comparison, and variant readings were noted from other editions of the two translations. Emmendations and other notes on the translations were given in the annotation to Kitagawa's Japanese translation Of the text, which was accompanied by Kitagawa's own Japanese commentary.55 The passages of Pramaijasamuccaya that were treated in this way were the fIrst half of the second, third, fourth and sixth chapters. In the fIrst half of each of these chapters DiIinaga offers his own views on the topics of reasoning for oneself, (svarthtinumana) demonstration, (pararthtinumana) the use of observed precedent (dr.stanta) in demonstration, and fallacy (jati) respectively. All this material was published in Kitagawa's 1965 monograph entitled Indo koten ronrigaku no kenkyu: Jinna (Dignaga) no taikei (A study of Indian classical logic: Dignaga's system), a corrected and expanded second edition of which appeared in 1973. This second edition also contained Kitagawa's English translation of a text by DiIinaga that is extant only in Yijing's Chinese translation with the title Qu yin jia she fun. 56 The Sanskrit title of this work was most probably Upadiiyaprajnaptiprakaraf)a. 57 Kitagawa regards it as a text that offers a synthesis of views of the Miidhyamaka school, the Yogaciira school and the Vaibh1i~ika school, and he expresses doubt whether it is truly a work by DiIinaga, since it contains views that appear to be inconsistent with views that DiIinaga expresses elsewhere. 58

26

Preliminaries

While Kitagawa Hidenori provided I apanese translations to key passages from the Pramtil;asamuccaya, the first English translation of a complete chapter of this work was provided by Hattori Masaaki, who published his translation of the first chapter of the work along with a truly critical edition of the Tibetan translations from which the English translation was made. In his introduction to the translation of the first chapter of that work Hattori offers a sketchy outline of the contents of the whole treatise in order to show how the first chapter fits into the plan of the work as a whole. Here Hattori emphasizes Diimaga's drawing a "radical distinction between the two kinds of objects--the particular (svalaksa-t;a) and the universal (stimtinya-lakSa-t;a). ,,5 He also states that Diilnaga was probably the first Indian logician to draw a distinction between reasoning for oneself} (svtirthtinumtina), which is "apprehension of an object through an inferential mark (liilga)," and demonstration (partirthtinumtina), which is the act of communicating to others what one has learned through reasoning for oneself. 60 About the apoha theory of meaning in Diilnaga Hattori says that according to Diilnaga's view a word indicates an object merely through the exclusion of other objects (anyapoha, -vyavrm). For example, the word "cow" simply means that the object is not a non-cow. As such, a word cannot denote anything real, whether it be an individual (vyakti), a universal (jati), or any other thing. The apprehension of an object by means of the exclusion of other objects is nothing but an inference. 61

Hattori notes the strong influence of DharmakIrti in Iinendrabuddhi, author of the only extant commentary on the PramtiIJasamuccaya, who appeals to such notions as that of causal efficiency, that is, the power to produce an effect, as the principal criterion of sensible objects which alone are fully real. Iinendrabuddhi explains Diilnaga in terms of this concept despite the fact that it is entirely absent in Diilnaga's work. 62 The reason that Diilnaga's work was all but forgotten after the time of DharmakIrti is according to Hattori because DharmakIrti's "PramtiIJavarttika is much richer in contents and more penetrative in arguments" than the PramtiIJasamuccaya and therefore took its place in the Buddhist academic world. 63 Apparently it is his view that DharmakIrti's work is an improvement upon Diilnaga's but that aside from these improvements and expansions the PramaIJavtirttika is essentially consistent with and faithful to the PramaIJasamuccaya. In keeping with this view Hattori draws heavily upon DharmakIrti and the later Buddhist epistemologists in his plentiful and most informative annotations to his translation of the first chapter of the PramaIJasamuccaya. As we shall see below in section 1.2.6, a number of the features of Hattori's approach that I have just outlined come in for criticism in the work of Radhika Herzberger.

Chapter One

27

Hattori's "Mlmamsaslokavarttika, apohavada-sh5 no kenkyu," published in two parts in 1973 and 1975, sheds light on the form and content of Diimaga's version of the apoha theory of meaning that Kumarua Bhatt;a criticized in a section of his Mfmiimsii§lokaviirttika. This study includes an appendix containing passages from the fifth chapter of the PramiifJasamuccaya that have been quoted or paraphrased in later Sanskrit works or that Diimaga himself quoted from the works of earlier authors. Although nearly half these passages had been reported before by Rangaswami Iyengar (1927), Hattori was able to find many new fragments by searching through Santarak~ita's Tattvasangraha, KamalaSIla's Tattvasangrahapafijikii and Uddyotakara's Nyiiyaviirttika. In subsequent work Hattori has devoted himself to tracing some of the roots of DiIi.naga's apoha theory of meaning. In 1977 he published an article in which he focused on antecedents to the theory found in works representative of the Sautrantika school of Buddhism, and in 1980 he published an article tracing some of DiIi.naga's ideas and terminology to Bhartrhari. In the first of these two articles Hattori observes that "unlike the other schools, the Bauddhas denied that a word has a direct reference to any real entity whether specific or universal. They maintained that the function of a word is nothing more than anyiipoha, i.e., the differentiation of an object from other things. ,,64 He also reiterates the view stated in Hattori 1968 in saying that DiIi.naga "made a sharp distinction between the particular (svalak$afJa) , which is real and directly perceived, and the universal (siimiinyalak$afJa), which is mentally constructed."65 Being an abstraction that is "conceptually constructed" through the process of differentiation, the apoha-universal has no objective reality, but it has the important features of a universal in that it is one, eternal and related to a plurality of individual entities.66 Hattori traces three different antecedents of DiIi.naga's formulation of the apoha theory of universals, two of which he acknowledges to have been pointed out by previous scholars. Following Kunjunni Raja (1963:193), who had noted a similarity between the Buddhist apoha theory and the view of the pre-Pataii.jali grammarian Vya(ii, Hattori discusses the different theories of meaning among the ancient grammarians: VyaQi, who held that the meaning of a word is any specific entity (dravya) of a class, set forth the view that the meaning of a sentence consists in the differentiation (bheda) of each specific entity from the others in the same class. On the other hand, it was maintained by Vajapyayana that a word refers directly to a universal (jali,) and that the meaning of a sentence is nothing other than the relation (samsargaJ of the universals which are denoted by different words in the sentence.6

28

Preliminaries

According to this theory advanced by Vya<;li, Hattori explains, in a sentence such as "Here is a white cow," the word "cow" applies promiscuously to any cow, but the word "white" serves to differentiate the cow in question from cows of any colour other than white; and while the word "white" refers promiscuously to anything that is white in colour, the word "cow" differentiates the white object in question from anything that is not a cow. This formulation is so much like what we find in Diimaga that we are safe in seeing VyiiQ.i as one of the forerunners of Diilnaga's apoha theory of meaning. But whereas in Vya<;li's theory the words in the sentence applied to universals as real entities, in Diilnaga's theory they refer only to mental constructs. A second forerunner of Diilnaga's theory, says Hattori, is Bhartrhari, who upheld a version of Vajapyayana's theory. In Bhartrhari's version of this theory, a universal's existing simultaneously in a plurality of distinct individuals implies that a word namines universals can apply to individuals only by disregarding their specificity. This, argues Hattori, is the basis of ,the Buddhist view that concepts are formed by disregarding the differences (bhedagraha) among perceived particulars, a view that plays an important role in later Buddhist apoha theorists but not in Diilnaga. But an aspect of Bhartrhari's thought that did find its way directly into Diilnaga's work is Bhartrhari's notion of what a sentence signifies, a connection that Hattori pursues in his 1980 article. Here it is explained that it was Bhartrhari's contention that the meaning of a sentence (vakyartha) is not merely the sum of the meanings of the individual words in the sentence but rather is an immediate flash of intuition (pratibhti) that arises when the whole sentence is heard. By comparing sentences against one another, one can analyse out common elements of meaning that may be assigned to word stems and grammatical suffixes, but the primary unit of meaning is the sentence rather than the word. Hattori shows that Diilnaga takes this theory directly into his discussion of sentential meaning in the fifth chapter of the Pramaf)asamuccaya. 69 A third antecedent that Hattori Masaaki sees to Diilnaga's apoha theory of meaning is its commitment to nominalism, which Hattori traces back to the Sautrantika concept of "nominal existence (prajfiapti-sat)." Here he points out that in the abhidharma system of the Vaibha~ika school of Buddhism, there was a category of dhannas that were said to be unrelated to thought (cittaviprayukta) and in this category were dhannas that were supposed to account for our understanding of the meanings of words and sentences. Vasubandhu, endorsing the view adopted by the Sautrantika school, on the other hand, denied the ultimate reality of this category of dhannas and relegated them to the status of merely nominal or

Chapter One

29

conceptual entities. 7o Given that Diimaga wrote a commentary on Vasubandhu's Abhidharmakosa and acknowledged his indebtedness to Vasubandhu in other ways, Hattori argues that it is likely that he was directly influenced by this Sautriintika view that the basis of our understanding of language is situated in pure ideas rather that in external realities. Many of the ideas that have been briefly outlined in the preceding discussion of Hattori's various contributions to the study of DiIinaga's thought will be dealt with in greater detail in later chapters of this work. Enough has been said here to show that Hattori made significant advances in the direction pointed out by Frauwallner, namely, in tracing DiIinaga's indebtedness to Bhartrhari and the earlier grammatical tradition as well as to the nominalism implicit in the abhidharma of the Sautrantika school. It will also be obvious as this work proceeds how much I am indebted to Hattori's scholarship despite my view that he underestimates the philosophical importance of the differences between DiIinaga and his later interpreters. A third Japanese scholar who has done a great deal to continue in the tradition of his senior colleagues Kitagawa Hidenori and Hattori Masaaki in shedding light on the thought of DiIinaga as Dmnaga himself expounded it has been Katsura Shoryu. Among his achievements has been a new Japanese translation of DiIinaga's Nyiiyamukha from the Chinese translations of Yijing and Xuanzang (entitled Yin ming zheng Ii men [un in Chinese).71 This study contains much information on parallel passages from Tibetan translations of other texts by DiIinaga as well as Sanskrit fragments. In addition to this study of the Nyiiyamukha Katsura has written a number of other studies on various aspects of DiIinaga's logic and epistemology, among which are his sketches of DiIinaga's apoha theory of meaning (1979b) and his discussion of the three features (trairiipya) of a legitimate inferential sign (Uliga) (1983 and 1986a). The fruit of many of these shorter studies has been a comprehensive monograph (1986b) in Japanese on the history of the concept of pervasion (vyiipti) in Indian logic from the time of Caraka to the time of DharmakIrti. This study traces the evolution of the principles of inference and demonstration in India from the time when these principles consisted in little more than guidelines in the conduct of debates and informal practical reasoning until the time when they developed into a somewhat more rigorous system of epistemology. The careful and thorough method of historical research employed by Katsura has added considerably to our understanding of DiIinaga's system of epistemology.

Preliminaries

30

1.2.6 Radhika Herzberger We have already seen that Frauwallner and Hattori both made advances in the study of the relationship between Diilllaga's theory of language and universals and Bhartrhari's theories on those matters. But as useful as their contributions were in showing the extent of Diilllaga's indebtedness to Bhartrhari, these two scholars still did little more on this particular question than to point the way that future research might take. The path that they pointed out was explored at great length by Radhika Herzberger in her study entitled Bhartrhari and the Buddhists: an essay in the development of fifth and sixth century Indian thought. This work focuses on three philosophers who had much to say on the related questions of language, universals and epistemology: Bhartrhari, Diilllaga and DharmakTrti. In her introduction to the work Herzberger states her central argument as follows: My thesis is that logical theory in India was an outgrowth of concerns with language. I argue that Indian logic had its origin in cross-currents between a grammarian, Bhartrhari, and a Buddhist philosopher, Dignaga, who took opposing positions on an aphorism concerninj names that had been enunciated many centuries earlier by Katyiiyana?

Not only does Herzberger correct for the oversight of those scholars preceding her who had neglected the importance of Bhartrhari as a factor in the development of Diilllaga's thought, but she also corrects for those scholars who have seen DharmakTrti as a faithful expositor of Dlllllaga's views. Indeed it is her contention that "in the debate between Bhartrhari and Dignaga, DharmakTrti, Dignaga's commentator, stood with Bhartrhari, against Dignaga. "73 She states this contention even more strongly in the conclusion to her work in which she states that while she has not provided an answer to why it is that Diilllaga's thought was superseded by the work of his commentator, "I have cast doubt on the verdict that India forgot Dignaga because what was essential in· his thought survived intact in DharmakTrti's. I have argued that DiIinaga's thought is not encomQassed by the greater depth of Dharmakmi' s, rather it is washed away by it."74 The grammatical dictum of Katyayana that holds a central place in the disputes among these three philosophers is, according to Herzberger, the fifth varttika on Pfu:1ini Siitra 5.1.119, the siitra that deals with the semantics of the suffixes -TVA and -TA that may be added to Sanskrit words to make abstract nouns. Herzberger translates Katyayana's varttika as follows:

Chapter One

31

[The occasioning basis for the use of a name is] that quality because of whose presence (bhiiva), a name is applied to a thing. [The addition to the nominal suffixes] tva or tal [is taught], in the signification of this quality?5

Herzberger also agrees with Hattori Masaaki that the debate between Vyat;ii and Vajapyayana over the question of what it is that a word names is in the background of the theories of language advanced by Bhartrhari and Diimaga. She observes that "against Vajapyayana, who had, in order to ensure their universal application, argued that Vedic injunctions should be interpreted as having reference to universals, Vyat;ii had mockingly observed that cows, not cow universals, are milked and goats, not goat universals, are sacrificed. "76 She argues that Bhartrhari tried to reconcile this dispute between the two ancient grammarians by appealing to a modified version of Katyayana's varttika. In this modified version, she says, the original "quality (gwJa)" that Katyayana had referred to is elevated by Bhartrhari to the status of a universal, which is bifurcated in two, namel~, into "a word universal (sabdajati) and a thing universal (arthajati)." 7 This theory of two universals allows Bhartrhari to hold the view that "a word signifies its own universal (sva jatO 'first of all' (prathamam) , it is then superimposed upon the universal belonging to things (arthjati)."78 In suggesting that the primary basis of applying words is a word universal that is superimposed by the mind upon thing universals, which are therefore not the real basis of applying words at all, Bhartrhrui is thought by Herzberger to be at odds with both Katyayana and Vyat;ii, for whom words were grounded more in external realities than in mind-borne abstractions. In reaction to this view, she says, Diimaga came to "chrunpion Vyat;li's cause,"79 by positing a theory whereby (a) names are given to spatio-temporal individuals on the basis of a spatio-temporal quality; (b) the content of a name does not 'exceed over' the content of the quality belonging to the individual which is named; (c) uniform application of names presupposes a uniform ground of application; (d) names have analytic and antonymic content and are, therefore, at some level, related to other names; (e) the qualifierqualificand relation can be accounted for without postulating ideal objects. 80

Drawing on a passage from Kitagawa Hidenori's English translation of the Upadiiyaprajfiaptiprakarar;a, in which Diimaga argues that persons and other continuants are neither purely illusory nor purely real but are ideas "constructed in dependence on realities," Herzberger claims that DiIinaga could 110ld the view that one and the same object possesses both universals and particularity. Rejecting the claim of Hattori Masaaki that there is for DiIinaga an unbridgeable gap between particulars and universals,

32

Preliminaries

Herzberger says "I believe that the universal (samanya) on the basis of which names are given to individuals is completely resident in the bearer... and is not cut off from its support ... .!t is not exclusively a property of names ... ,nor is it the product of a beginningless,habit-energy (vasana), as DharmakIrti would have it. "81 But, whereas Dirinaga had been content to believe that words could apply to objects on the basis of universals that resided in the objects themselves, Herzberger argues that DharmakIrti returned to a position rather like Bharrrhari's, according to which "objects are internal and projected onto an external sphere."82 Herzberger's arguments are complex in detail and subtle in execution, and they are bound to suffer in this brief synopsis that has been given here. Enough has perhaps been given though to show the main lines of her position and how it differs from the characterizations of the thought of both Diimaga and DharmakIrti that are to be found in the scholarship that preceded her. Of particular importance is her demonstration of the extent to which Dirinaga's theories "were addressed to the more universal context of those theories [of language] which constituted the common subject matter of Buddhists and non-Buddhists," rather than to "the narrow sectarian interests of Buddhism alone,"83 and her claim that Dharmaklrti "washed away" the important features of Dirinaga's thought by replacing them with theories of his own that were supposedly immune from the criticisms to which Dirinaga's thoughts had been subject. We shall have occasion to return to a number of issues that she raises and to discuss them more thoroughly in later parts of this work.

1.3 The argument of this book In the above review of what modem scholars have had to say about Dirinaga's logic and theory of language, we have seen that some have held the view that the fundamental preoccupation of Dirinaga and his followers was metaphysical in nature. Representatives of this view are Satkari Mookerjee, Dharmendra Nath Shastri and Raja Ram Dravid, all of whom in one way or another see the Buddhist logicians as participants in the sectarian polemics that characterized much of Indian intellectual life in the medieval period when Buddhism and Hinduism were vying for supremacy. In contrast to this view Erich Frauwallner and A.K. Warder have seen epistemology as being of central importance to Dirinaga, arguing that whatever metaphysical stances the Buddhists took were deIivative from their basic epistemological observations. Radhika Herzberger, on the other

Chapter One

33

hand, has argued that the principal concern of Dirinaga was with language and that it was out of his philosophy of language that his logic and epistemology evolved. Apparently (and perhaps only apparently rather than actually) there is a lack of consensus among scholars on what Dirinaga's principal interests were in writing his treatises on logic, language and the general question of the interpretation of signs. My main interest is not to try to reconcile all the apparent differences in emphasis that have arisen in previous scholarship, but the portrayal of the Buddhist epistemologists that I present in the remainder of this book may in fact be construed as presenting a view that if correct would render much of the disputes among previous scholars somewhat irrelevant in the sense that it might show that several of them have missed the point of Dirinaga's philosophy altogether. It is my contention that Vetter and Steinkellner came very close to the heart of the matter in the claim cited above (p. 5) that "The Buddhist's actions are oriented towards the goal of emancipation," for I shall argue that it is the quest for nirvii1:za that most interests Dirinaga. I also agree with their claim that in the philosophy of Dharmaldrti the connection between epistemology and the quest for nirvii.!:za is the preoccupation with providing a rational basis for Buddhist dogmatics as based upon the Buddha's words and that this preoccupation was grounded in a recognition that it was the Buddha who had shown the path to nirvii.!:za but that the Buddha's teachings could not be accepted on blind faith. I do not, however, believe that Dirinaga's interest in epistemology is motivated by quite the same considerations as is Dharmakirti's. Dharmakirti is undeniably a polemicist with a strong interest in demonstrating the truth of the Buddhist position through rational argument. He is fIrst and foremost a Buddhist apologist. But it is not, I shall argue, Dirinaga's purpose to demonstrate that anything is the case; his chief motivations were not, in other words, polemical in nature. Rather, I shall argue, his writings on logic and epistemology should be regarded as the contents of his own meditations directed at attaining nirviil)a directly instead of merely arguing about how nirviil)a is to be attained. In making my case that Dirinaga's principal interest was in nirviil)a, I shall have to clarify just what view of nirviil)a was current in his thought. Unfortunately a great deal of confusion about Buddhist nirviil)a still prevails in modern scholarship, where it has so often been treated as a mystical accomplishment of some kind. Why this misconception persists is a question of great complexity that there is no need to investigate in any detail here. But let me try to offer an abbreviated answer to the question. The context in which the virulent misconception of nirviil)a as a mystical

34

Preliminaries

state is most kept alive and robust is probably in the comparative study of religions, where there is often an underlying assumption that there is such a thing as religion in general and that particular religions all somehow share in this common religiosity. Among the practitioners of the dubious science of comparative religions there are still many who are in sympathy with the ideas of such pioneers in the field as Rudolf Otto, who sought to find the essence of religiosity in the concepts of the numen and the numinous. Otto managed to find what he called the "exuberant nature of the numen" even in Buddhist nirvii1)a. He writes: Nirvana.. .is only in appearance a cold and negative state. It is only conceptually that 'Nirvana' is a negation; it is felt in consciousness as in the strongest degree positive; it exercises a 'fascination' by which its votaries are as much carried away as are the Hindu or the Christian by the corresponding objects of their worship. I recall vividly a conversation I had with a Buddhist monk. He had been putting before me methodically and pertinaciously the arguments for the Buddhist 'theology of negation', the doctrine of Anatman and 'entire emptiness'. When he had made an end, I asked him, what then Nirvana itself is; and after a long pause came at last the single answer, low and restrained: 'Bliss--unspeakable'. And the hushed restraint of that answer, the solemnity of his voice, demeanour, and gesture, made more clear what was meant than the words themselves. 84

The notion of nirv(1)a that Otto acquired by interpreting the "body language" of one Buddhist monk and by putting more confidence in this implicit sign than in the monk's explicit words has been perpetuated in various forms and is still fairly common a,mong well-meaning Christians who try hard to find some common ground between their own religion and Buddhism. But as convenient as it might be for those who practise "interfaith dialogue" if nirv(1)a were to tum out to be simply another version of God that the Buddhist votary could worshi~ without quite understanding that he is after all an "anonymous Christian," 5 a close examination of what Buddhists themselves have consistently written about nirv(1)a will not support such a view. It simply will not do, I believe, to continue past habits of trying to discuss Buddhist nirvii1)a in such terms as "soteriology," a term that implies the aid of a saviour (soter) and makes no sense at all in most Buddhist contexts, or "mysticism" or "gnosis," terms that imply the existence of a type of knowledge that transcends both the empirical and the intellectual and thus also fail to apply very well to most Buddhist contexts. My task, therefore, is to find the proper language in which to discuss Buddhist nirv(1)a, which amounts in part to findi~g a group of philosophers who are familiar to speakers of European languages and who pursued a goal reasonably similar to the Buddhist goal of nirv(1)a. I shall

Chapter One

35

argue in chapter two that such philosophers can be found in the Hellenistic philosophers who came to be called Skeptics. I shall very briefly describe their religious goal and their methods of achieving that goal and shall compare them with clearly discernible trends to be found in both Canonical Buddhism and in the early Mahayana Buddhism of Nagarjuna. It is argued that in both early Buddhism and in the Skeptics one can find the view put forward that man's pursuit of happiness, the highest good, is obstructed by his tenacity in holding ungrounded and unnecessary opinions about all manner of things. Much of Buddhist philosophy, I shall argue, can be seen as an attempt to break this habit of holding on to opinions. Even the polemical side of some Buddhist philosophers can be viewed not as an exercise in stubborn sectarian dogmatics but rather as a steadfast rejection of all pretenses to having access to anything that transcends the human condition that we all know and hate. Buddhist nirviil;a, then, is seen not as a transcendence but as an acceptance of our own limitations and frailties and ignorance. Just as nirviil)a is a conquest of death only in the sense that it is an acceptance of death and therefore a conquest of our fear of dying, nirviil)a is also a conquest of ignorance only in the sense that it is an acceptance of the limitations .on our knowledge and therefore a conquest of our fear of being in the dark. Much of our opinionated tenacity is grounded in an unwarranted confidence in our own language and in the concepts that our language expresses. Showing that this confidence is unwarranted results in a kind of nominalism. In chapter three I shall try to show that nominalism is a prominent feature of Buddhist thought from the early Canonical period right through to the time of Diimaga and especially that it is an important aspect of the philosophers who influenced Dinnaga. Seen as a corollary of Buddhist skepticism, which is in tum seen as essential to the quest for nirviil)a, nominalism ceases to be inconsistent with the basic principles of Buddhist logic. I shall argue in chapter four that logic for Dmnaga is not intended to serve as a means of adding to our knowledge but rather as a means of subtracting from our opinions. The criteria of inferential certainty that DiIinaga puts forward are very· strict, so strict that very few of our opinions can measure up to them. Whereas DharmalcIrti tended to see this snictness as a weakness in Dinnaga's logic, since it was in DiIinaga's system virtually impossible to arrive at any sort of reasonable inductive certainty, I shall argue that the strictness of Dmnaga's criteria was no oversight on his part but was quite deliberate, for it served his overall skepticism quite· well. In chapter five I shall argue that DiIinaga's nominalism was a corollary of his skepticism in much the same way that nominalism was a corollary of skepticism ill Buddhism generally. In these two chapters I shall also try to show that skepticism and nominalism were

36

Preliminaries

trends in those of DiIinaga's works that are devoted to logic and epistemology. Following these chapters I shall provide an annotated translation of parts of chapters two and five of DiIinaga's PramiilJasamucca:ya.

-- Notes--

Chapter 1. Preliminaries 1. Stcherbatsky 1932:39-46. 2. Stcherbatsky 1932:40. 3. Stcherbatsky 1932:46. 4. Stcherbatsky 1932:46. 5. Stcherbatsky 1932:37. 6. Van der Kuijp 1979:6. 7. Conze 1962:264. 8. Conze 1962:265. 9. Wayman 1958:40. 10. Conze 1962:10. 11. Steinkellner 1982: 11. 12. Hayes 1984. i3. Warder 1971:1. 14. Warder 1970:448. 15. A complete list of Dinniiga's works appears in Hattori 1968:6-11. 16. Malvania 1967:390. 17. J. W. de J ong (1986: 128) argues that it is a mistake not only to rely too heavily on later sources when piecing together the history of Indian Buddhist thought, but also to try to

Chapter One

37

study later Tibetan thought without a profound knowledge of its Indian antecedents. His observation is so obviously true that it is a little surprising that relatively few modem scholars have done research directly with Ditmaga's works in their proper historical context. 18. Vidyabhu$ll!Ja 1905:218. 19. "Chenna" is how the Chinese characters used to transliterate Diimaga's name are rendered in the modem pinyin system of romanization. The Wade-Giles rendering would be Ch'en-na. The form of romanization that VidyabhU~al)a uses is "Jina," which he says that Japanese scholars mistakenly took to be a transliteration of the Sanskrit word ''jina'' meaning conqueror, which is frequently used as an epithet of the Buddha. See Vidyabhu~al)a 1921:272. Perhaps VidyabhU$ll!Ja's remark derives from the fact that the Japanese pronunciation of the two characters used to render Ditmaga's name into Chinese is Jinna. 20. See Vidyabhu$al)a 1921:277-288. 21.

VidyabhU~ll!Ja

1909:78-101.

22. Stcherbatsky 1932: 10. 23. Stcherbatsky 1932:12. 24. Stcherbatsky 1932:24. 25. Stcherbatsky 1932:26-27. 26. Stcherbatsky describes the realm of particulars in 1932:79-118. Of especial importance are pp: 81-86 and pp. 106-108. 27. Stcherbatsky's outline of the realm of judgement, concepts and other related matters is given its most precise formulation in 1932:204-220. 28. Stcherbatsky 1932:181-190. 29. Mookerjee 1935:xXxv. 30. Mookerjee 1935:xliv. 31. My position on this matter is outlined more fully in Hayes 1984. 32. Mookerjee 1935:xxxviii. 33. Mookerjee 1935:xxxvii. 34. Mookerjee 1935:100. 35. Mookerjee 1935:106. 36. Mookerjee 1935:108 and 109. Frauwallner (1935:101) reports that this argument was used by DharmakIrti.

38

Notes

37. Mookerjee 1935:108. 38. Mookerjee 1935:116. 39. Mookerjee 1935:116. 40. Mookerjee 1935:117. 41. Dravid 1972:339. 42. These articles appeared in the following issues of Wiener Zeitschriftfur die Kunde des Morgenlandes: Vol. 37 (1930) pp. 259-283; Vol. 39 (1932) pp. 247-285; Vol. 40 (1933) pp. 51-94; Vol. 42 (1935) pp. 93-102; Vol. 44 (1937) pp. 233-287. 43. The numbering of the verses in modern editions of Sanskrit texts is often problematic, and it is not unusual to find different numberings by different editors of the same text. Thus verses 42-187 in Frauwallner's edition correspond to verses 40-185 in Gnoli's 1960 edition of the Sanskrit text of the entire svarthanumana chapter of PramiilJaviirttika. 44. Frauwallner 1930:259. 45. See Frauwallner 1935:93 and 101. In the latter place this doctrine is spelled out especially clearly: n ••• die Einzelding augenblichlich sind und sofort nach ihrem Enstehen wieder vergehen ... n (particulars are momentary and disappear again immediately after their coming into being.) 46. This is mentioned in Frauwallner 1935:93 and discussed in the fIrst half of the article, that is, up to p. 97. 47. Frauwallner 1935:95-96. 48. Frauwallner 1959b:135-136. 49. Frauwallner 1959b:107-1l6 and 145-152. Further discussions of Bhartrhari's influence on Diimaga are to be found in Hattori 1977:50-52 and R. Herzberger 1986. See section 1.2.6 below. 50. Frauwallner 1959b:103. 51. Frauwallner 1959b:101-103. 52. Frauwallner 1959b:l04-105. 53. Frauwallner 1959b:105. 54. Throughout this work I shall follow the custom of presenting the names of Japanese authors in the order in which their names are given in Japanese, with the surname first and the given name second. The one exception to this rule is the name of D.T. Suzuki, whose name is so well-known to the English-reading world in its Western-style orde. that it would be confusing to alter it.

Chapter One

39

55. A rough draft of Kitagawa's English translation of these passages was prepared for a seminar on Buddhist epistemology when Kitagawa was a guest professor at University of Toronto in the 1970's. Unfortunately Prof. Kitagawa died before he was able to polish the English translation enough for the purposes of publication. A photocopy of this translation was made available to me by Elizabeth Hayes, who participated as a student in the Toronto seminar. 56. In the Wade-Giles system of romanization, the title is Ch'ii-yin-chia-she-Iun, and the translator's name is I-chirig. So they appear in Kitagawa 1965 and in Hattori 1968. 57. Hattori 1968:8. 58. Kitagawa 1965:436. 59. Hattori 1968: 12. 60. Hattori 1968:12. 61. Hattori 1968: 12-13. 62. Hattori 1968:14. 63. Hattori 1968:15. 64. Hattori 1977:47. 65. Hattori 1978:47. According to Hattori an example of a particular for Ditmaga would be an individual such as a particular cow, which is "an indivisible unity of various aspects." Whether or not this observation is accurate is a question that should be left open for the time being. I shall present an alternative account of what a svalak.$afJa might be below in chapter five. 66. Hattori 1977:48. 67. Hattori 1977:49. 68. Hattori 1977:50. 69. Hattori 1980:61-64. 70. Hattori 1977:52. 7l. Katsura 1977, 1978, 1979a, 1981, 1982 and 1984. 72. R. Herzberger 1986:xvii. 73. R. Herzberger 1986:xviii. 74. R. F1erzberger 1986:241. 75. R. Herzberger 1986:24. The Sanskrit for this varttika is given, along with a translation and brief discussion, in Hayes (1983:715 n.20). The Sanskrit reads: "Yasya glll}asya

40

Notes

bhi'ivad dravye sabdanivesas tadabhidhane tvatalau." My translation: "Th~ suffixes -tva and -ta are used to denote that characteristic on account of the presence of which in an object a given word is applied to it." 76. R. Herzberger 1986:110. 77. R. Herzberger 1986:110. 78. R. Herzberger 1986:30-31. 79. R. Herzberger 1986:111. 80. R. Herzberger 1986:112. 81. R. Herzberger 1986:115. 82. R. Herzberger 1986:213. 83. R. Herzberger 1986:106. 84. Otto 1923:38-39. 85. Knitter (1985:128-131) offers a neat summary of Karl Rahner's concept of the anonymous Christian, the purpose of which is in Knitter's words "to broaden and engender more 'optimistic' Christian attitudes toward other believers. In showing that other believers can be called 'Christians without a name; Rahner tries to break through Christian exclusivislU."

Chapter 2 Rational Skepticism in Pre- Dinnagan Buddhism What I shall try to show in the present chapter and in the following chapter is that there are two currents present in Buddhist canonical literature from the very beginning and that these two currents are also predominant in the philosophical writings of early Mahayana systematizers such as Nagarjuna and Vasubandhu. The first of these currents is a kind of skeptical rationalism according to which there is no Imowledge aside from that which meets the test of logical consistency, and moreover very few of our beliefs meet this test. And the second is a form 'of nominalism, according to which universals are not features of the external world that exist independently of our awareness of the world but rather are born of the attempts of awareness itself to organize the data of sense experience. In later chapters I shall be trying to show that these two currents are the principal currents in Dirinaga's system of thought as well. My motive in showing this is not to support the conclusion that Dirinaga had nothing new to say on the these issues or that he was doing nothing more than drawing out in full detail the implications of earlier Buddhist dogmatics. Nor is my motive to deny that Dirinaga was heavily influenced by non-Buddhist sources, for I am convinced by the evidence adduced by Frauwallner and by R. Herzberger that Dmnaga did indeed owe a great debt to Bhartrhari and to the whole tradition of the Sanslait grammarians. Rather my motive in presenting the evidence of the present chapter is simply to show that in holding the ideas that he held and in presenting the kinds of arguments that he presented, Dirinaga was taking a radical departure neither from the course of early canonical Buddhism nor from early Madhyamaka Buddhism.

41

42

Chapter Two

2.1 The foundation of skepticism in the Nikayas The corpus of Buddhist canonical literature is so vast and diversified that it is possible to find in it, even if only in the form of a vague anticipation, an antecedent to nearly every intellectual trend that occurred in later scholastic Buddhism. This statement is certainly true if we regard as canonical all the various Mahayana sutras that came into existence in the early centuries of the Common Era. But it is only slightly less accurate to say of the Pali Canon that it contains anticipations of the currents of much of later scholastic Buddhism, even of that which is based explicitly on later Mahayana writings, for the novelty of the Mahayana sutras lay principally in a great proliferation of strikingly new and more imaginative ways of stating themes that are, despite the novelty in expression, for the most part the same as those found in the corpus of the PaIi Canon. Owing to the fact that the Nikaya and A.gama literature was compiled over a period of time, it is better not to try to regard that corpus of literature as an historical document of the life and teachings of Siddhartha Gautama (in PaIi, Siddhattha Gotama) alias the Buddha Sakyamuni and his immediate followers, but rather as a large literary creation, which is fictional but based loosely on historical events, and in which Sakyamuni Buddha appears as a persona. Moreover, given that the Nikayas reflect the predispositions of a large number of compilers, it is not surprising that one can in fact find several different personae who bear the name Sakyamuni. Therefore, anyone who sets out to discuss the teachings of the Buddha as outlined in the Nikaya literature does so only at the risk of being accused of parochialism. Bearing that in mind, I wish to emphasize here that in the discussion of Nikaya literature that follows my only claim is that a persona appears there, one of several who bears the name of Salcyamuni Buddha, who distinguishes himself from other philosophers of his time by being wary of holding unnecessary opinions and by warning that judgements must be constantly questioned. It is well known that the aim of Buddhist philosophy as it appears in the agama literature is to provide a theoretical underpinning for Buddhist practice and that the aim of Buddhist practice is to achieve nirva1].a. NirvliI).a is the eradication of various sources of torment and distress, chief among which is the set of errors in the understanding that are known collectively as avijja or misconception. Errors of understanding can take many forms and are generally the product of what the Buddhists called "unprincipled thinking (ayoniso manasikara)," which is the failure to think

Rational Skepticism in Pre-Diimagan Buddhism

43

matters through thoroughly and precisely in accordance with natural principles.! Such carelessness in thinking results either in inconect opinions or in opinions that are simply ungrounded. An unsupported view is called ditrhi in Pali and drsri in Sanskrit. References to the hazards of ditrhi abound in the Pali Canon, so rather than being exhaustive in citing references to these hazards, let me simply offer a few representative samples along with an interpretation of them.

2.1.1 The Sutta Nipata Along with the Dhammapada, the Sutta Nipata is one of the most basic texts in all of Buddhism in that it is usually one of the first to be learned by anyone taking instruction in Buddhism and in that it contains the simplest expressions of basic attitudes that are treated more systematically and at greater length elsewhere. The Paramatrhakasutta of the Sutta Nipata comprises the following eight verses. 796. A person who persists in opinions regards as a waste everything other than that which, thinking "it is supreme," he regards as best in the world. Therefore he fails to get beyond disputes. 2 797. Then grasping at just that which he sees as commendable to himself in rules of conduct and vows and in what is seen, heard or thought, he regards everything else as a loss.3 798. The experts caIl that thing a shackle owing to which one considers all else a waste. Therefore the monk should not pursue rules of conduct and vows and what is.seen, heard or thought. 4 . 799. Nor should he form an opinion of people either through knowledge or through rules of conduct and vows. Nor should he present himself as an equal, nor should he think of himself as mean or excellent.5 800. Giving up assumptions and not taking them up again, he does not pursue even knowledge. Indeed he does not side with any party in controversies, nor does he believe any opinion whatsoever. 6 801. He who has no inclination for either extreme for being born or not being born either in this world or in another has no attachment whatsoever that is grasped after discriminating among properties? 802 .. Here not even the slightest judgement is formed by him with respect to what has been seen, heard or thought. How could anyone in this world doubt that unassuming brahman?& 803. They do not form jUdgements, they do not show reverence; not even virtues are accepted by them. Brahmans are not to be guided by rules of conduct and vows. Having gone elsewhere, such a one does not retum.9

44

Chapter Two

The sentiment that runs through these verses and many others like them in the Sutta Nipata is clearly that opinions lead to strife and disharmony among people. 1O People become attached to their own opinions and their own doctrines and come to disdain people who hold other opinions and doctrines. People form hierarchies of doctrinal acceptability, ranking some doctrines and the people who hold them above other doctrines and the people who hold them. It might be noted in passing that this principle of ranking doctrines is evident in many of the philosophical primers that were produced in later times in India. For example the structure of such works as the fourteenth century Sarvadarsanasamgraha of VidyaraI).ya alias Madhava is to present fIrst those opinions that the author regards as least valuable and to end with those that he regards as closest to the truth. Madhava, incidentally, places Buddhism very near the beginning of his work, considering it only slightly better than Carvaka materialism. The practice of ranking doctrinal systems hierarchically also became common in later Buddhist authors who in their introductions to philosophy often distinguished between some systems of Buddhist thought that were said to be aimed at people of lesser intelligence and other systems that were said to be for the more sophisticated. A typical feature of all treatments of philosophy that proceed in this way is that in their compendious treatments of opinions of "lesser value" than their own, they present caricatures rather than accurate. portraits of the rival philosophical systems. A number of modem scholars, relying on such compendia for their information on philosophical systems that were not favoured by the authors of the compendia, have been led into presenting rather badly distorted accounts of the conclusions and arguments of philosophers who were far more subtle than their opponents were willing to acknowledge. Such distorted accounts, and the attitudes of contempt that so often accompany them, provide good instances of the kind of unsavory consequences that the Sutta Nipata says come of the pursuit of dirrhi or opinion. The attitude towards the value of opinions expressed by the author or authors of the Atthakavagga (Section of Octads) of the Sutta Nipata is that opinions are shackles that impede one's attainment of peace of mind. The true sage, the Buddha, on the other hand, has forsaken all opinions and has given up all disputes and in so doing has attained inner peace. 837. The Lord said "Magandiya, there is nothing that he accepts, saying 'I assert this!' after discriminating among properties. And looking among the opinions without assuming any, seeking inner peace I found it." 11 "

45

Rational Skepticism in Pre-Dirinagan Buddhism A little further on in that same discourse, the Buddha says: 847. "One who is free of judgements has no shackles. One who is set free by wisdom has no delusions. But those who take up judgements and opinion go about in the world being contentious. ,,12

.

In this expression of the Buddha's attitudes towards opinions, there are

some matters that need to be explained. The attitude is not necessarily that of a skeptic, who doubts whether it is possible to attain certainty about a given range of questions. Rather it is the expression of an observation that the pursuit of truths leads to internal turmoil and that the conviction that one has found truths leads to social disharmony. People who claim to know the truth tend to be obnoxious in their penchant for pointing out the stupidity in those who hold alternative opinions. What still needs to be explained is whether this attitude that is attributed to the Buddha is supposed to extend to all opinions or only to some specified range of opinions. Buddhist tradition has not been univocal in its answer to this question. In some interpreters there is a tendency to class as opinion (ditrhi, dr$(i) only a certain prescribed set of opinions that are at odds with accepted Buddhist dogmas. On this interpretation, when the Buddha says it is salubrious to abandon all opinions, he means to abandon all false opinions (micchii ditthi). Stock examples of false opinions that already occur in the Nikaya literature are: that events happen adventitiously, or through divine intervention, or as a result of purely material causes independently of consciousness or will, as opposed to the accepted Buddhist view that they are the result of karma; that a person exists as an ultimate reality, as opposed to the accepted Buddhist view that persons are mere constructs of the mind; and that an existing person either ceases to exist upon the death of the physical body or continues to exist in some other form after death. But there also exists a more radical interpretation according to which when the Buddha prescribes the abandoning of opinions, he means the abandoning of all opinions on any issue whatsoever that do not have a direct bearing on the attainment of personal happiness and inner calm. Advocates of this latter interpretation can find support for their opinions in such Nikaya passages as the Brahmajiilasutta of the DIgha Nikaya and Mahavagga 65 of the Ariguttara Nikaya, tikanipata. Let us turn now to an examination of those two suttas. 2.1.2 Digha Nikaya: The Brahmajiila sutta The Brahmajiila Sutta of the DIgha Nikaya holds a central place in that collection of suttas in that as the first sutta in the collection it contains formulae that are repeated, with occasional alterations, in most of the other

46

Chapter Two

suttas in the collection. A long discourse, the Brahmajii/a cSutta also contains a great many of the doctrines and attitudes that the early Buddhists felt were central to their way of life. The circumstances of the sutta are described in the opening paragraphs as an occasion in which the Buddha was travelling with a number of monks between Rajagaha and Nalanda, and in their company was one wandering renunciate (paribbajaka) named Suppiya, who was proclaiming the Buddha's faults. Also in their company was Suppiya's disciple Brahmadatta, who was proclaiming the Buddha's virtues. The' Buddhist monks in the party grew concerned about this controversy between two non-Buddhists over the merits of the Buddha. Becoming aware of this concern in his followers, Gautama Buddha then gives a long address to the monks in which he first warns them not to become angry and displeased when non-Buddhists find fault with him or with his teachings, nor to become pleased and gladdened when outsiders praise him or his teachings. For reacting with feelings of either pleasure or displeasure will impair one's ability to assess whether what is being said of the Buddha is said accurately or inaccurately. So rather than feeling anger or joy the monks should remain calm of mind and dispassionately assess the merit of what has been said about the Buddha or his teachings or his community of followers. Following this general advice, the Buddha then tells his monks what kinds of things it would be accurate to say of him and of his teachings and how it is that his teachings and claims differ from those of other teachers. Other teachers, says the Buddha, have opinions about a good many things, and they teach these opinions to their disciples. But the Tathiigata has a superior sort of knowledge (abhififia), that is, an understanding that surpasses that of those teachers who are purveyors of mere opinion. "There are, monks, other properties that are profound, difficult to observe, hard to understand, peaceful, excellent, outside the realm of speculation, subtle, but understandable to those who are clever, which the Tathiigata, having himself experienced superior knowledge directly, makes known, and on account of which those who speak correctly should proclaim the Tathiigata's character as it really is. But what are those properties that are profound etc.? ,,13

Having asked this enticing question, Gotama Buddha then goes on to enumerate all the various things about which other teachers have opinions to which the Tathiigata's understanding is superior. Those opinions concern such matters as whether things have eternal essences that remain unchanging as the things themselves undergo changes of fonn and changes of accidental property; whether there is an unchanging creator whose

Rational Skepticism in Pre-Diimagan Buddhism

47

creatures are subject to change; whether or not the universe is infinite; and whether or not it is possible to know what is morally good. Teachers who hold opinions on matters such as those purport to back up their opinions in various ways. Some, says the Buddha, go into trances in which they supposedly learn things that ordinary people cannot know. Others rely on empirical observations of some part of the world and then generalize on those limited observations. Others appeal to a priori principles from which they derive their conclusions. But the Tathagata differs from all these other teachers in that he, unlike them, is aware that "these dogmatic tenets thus taken up and thus embraced will lead to such and such consequences and will lead to such and such a destiny." 14 If the results of holding opinions were salubrious, the reader is left to conclude, then the Tathagata would encourage his disciples ro have opinions on these matters. But in fact the Tathagata has no opinion on these matters whatsoever and does not encourage anyone else to have opinions on them, because holding opinions on these matters is like being trapped in a fine-meshed net cast by a clever fisherman; once one is caught in the net, one flops this way and that, but one never gets free. 15 The Tathagata is also said to stand clear of several opinions concerning the nature of nirvtil;a. In particular, the Tathagata is said to avoid the opinion held by some that nirvtif}a consists in indulging oneself in sensual pleasures, or that it consists in putting aside sensual pleasures and following purely intellectual pursuits I , or that it consists in putting aside intellectual pursuits and abiding in a state of blissful calm, or that it consists in giving up the attachment to bliss and abiding in a state of purely impartial serenity, or that it consists in attaining such a state of pure abstraction that one feels nothing whatsoever. As in the case with the various opinions concerning the nature and extent of the cosmos, the Buddha states that he, unlike other teachers, is aware that "these dogmatic tenets thus taken up and thus embraced will lead to such and such consequences and will lead to such and such a destiny." But holding such opinions about nirvtif}a will lead to nothing good; these opinions, too, are a snare that a person wouid be wise to avoid. If the Tathagata holds no opinions about the ultimate nature and extent of the cosmos and no opinions about the ultimate in human happiness, then it would appear that very little is left over about which to form an opinion. Moreover, one would be justified in wondering just what it is that this "superior knowledge (abhililiti)" that the Buddha claims to have could possibly be about. The answer to this question is given at Brahmajtila Sutta 2:29, immediately after the formula in which it is said

48

Chapter Two

that the Tathagata is aware of the consequences of holding various dogmatic tenets. He then goes on to say: The Tathligata understands that, and he also understands something better than that; and understanding that, he does no't become attached; and not being attached, he has found tranquillity all by himself; after realizing just as they really are the arising and the perishing and the sweetness and the peril and the abandoning of feelings, not clinging to them the Tathligata is, monks, set free. Just these, monks, are those other properties that are profound, difficult to observe, hard to understand, peaceful, excellent, outside the realm of specUlation, subtle, but understandable to.those who are clever, which the Tathligata, having himself experienced superior knowledge directly, makes known, and on account of which those who sReak correctly should proclaim the Tathligata's character as it really is.

In an era in which various teachers are gathering disciples around them

and making claims of supernatural powers and access to cosmic information that is beyond the ken of ordinary mortals, the superior knowledge of the Tathagata consists in no more than a full awareness of his own feelings (vedanii) and the realization that tranquillity is possible only by giving up being attached to them. 2.1.3 AIiguttara Nikaya: The Kesaputtas

Further evidence of an early Buddhist wariness of the hubris that so often accompanies claims to knowledge is found in the AIiguttara Nikaya in a short sutta that has been made famous by the importance given it by Walpola Rahula, who makes the somewhat exaggerated claim for it that it contains a piece of advice that is "unique in the history of religions.,,18 In this episode Gotama Buddha fmds himself in a district of Kosala called Kesaputta, where a group of people known as the Kaliimas live. On arriving in Kesaputta with a large number of monks, the Buddha discovers that his reputation has preceded him, for the Kalfunas have heard that he is a worthy, a fully awakened person perfected in knowledge and in practice, one who is well known, who understands people, who is an unsurpassed tamer of the human beast, a teacher of .luminaries and ordinary people, an illustrious buddha. Having himself experienced superior knowledge directly, he makes known this world with its luminaries, its death and its divinities to ascetics and priests, beings, luminaries and ordinary people. 19

The Kalfunas think it would be a good idea to visit such an illustrious person, so they approach him with a problem that has been troubling them for some time. Other teachers, they say, have been througH their district, but each in extolling the merits of his own opinions levels harsh criticisms

Rational Skepticism in Pre-DiIinagan Buddhism

49

against the teachings of others. As a result the Kalamas have become confused and no longer know whom to believe. It is in this context that the Buddha offers his celebrated piece of advice: It is appropriate indeed to be uncertain, Kalamas; it is appropriate to be . skeptical. For when something about which one ought to be uncertain is present, skepticism does arise. Do not, Kalamas, arrive at conclusions owing to hearsay, owing to tradition, owing to romour, owing to distinction in canonical works, on account of speculation, on account of methodical reasoning, owing to a study of appearances, after contemplation and acquiescing to an opinion, because of plausibility, nor by thinking "the ascetic is our revered teacher." But when, Kalamas, you yourselves know: "These properties are unhealthy, these properties are shameful, these properties are reproached by the wise, these properties when fulfilled and undertaken lead to harm and distress," then you must reject them. 20

After giving the Kalamas that general piece of advice, the Buddha invites them to consider their own experiences with such characteristics as selfish desire, anger and malevolence, asking them whether these characteristics have led to healthy results or harmful results. The implication is that nothing else really matters; all the other cosmological speculations over which teachers have disputes and because of which they revile one another's doctrines are superfluous. There is nothing that requires to be known beyond this simple piece of advice, said to be the core teachings of not only Siddhattha Gotama but of all who have awakened: "Not doing any wrongful deeds, undertaking what is wholesome, purifying one's thought. This is the discipline of them who are awakened. ,,2 The Tathagata's superior knowledge, remarkable only in its being so ordinary, is reminiscent of Socrates's account of the nature of his own wisdom, which he came to understand after examining a man reputed for his great knowledge: "However, I reflected as I walked away, Well,I am certainly wiser than this man. It is only too likely that neither of us has any knowledge to boast of, but he thinks that he knows something which he does not know, whereas I am quite conscious of my ignorance. At any rate it seems that I am wiser than he is to this small extent, that I do not think that I know what I do not know. ,,22

Like Socrates, Gotama Buddha exposes the sham of religious leaders of his day. The wisdom of brahmans who know the Vedas by heart but fail to live lives of pure conduct is likened to a barren desert (fril;a), ajungle (vipina), and a calamity (vyasana).23 Moreover, in claiming to lead others to a goal that they themselves have never realized, namely first-hand

50

Chapter Two

knowledge of Brahma, these brahmans, says the Buddha, ,teach of a religious goal that is in fact laughable (hassaka), vain (rittaka) and empty (tucchaka).24 In contrast to the three-fold knowledge possessed by brahmans, which consists of knowing the three collections of Vedic wisdom allegedly revealed to man through the R~is, Gautama Buddha's three-fold knowledge is said to consist of 1) knowing the details of his own former lives, 2) knowing that the present circumstances of other beings is a consequence of the fruitions of their own past actions, and 3) knowing the nature of distress (dukkha), how it comes into being, its cessation and how to achieve its cessation. 25 Since his knowledge and understanding is limited to just these matters, people should be careful not to exaggerate the extent of his wisdom. He warns, for example: They do not speak of me in accordance with what I have said, and indeed they slander me with untruth and falsehood, who say this: "The ascetic Gotama is omniscient and sees all things; he claims unlimited knowledge and understanding, as he says 'Whether walking, standing, sleeping or awake, I have knowledge and understanding continually and perpetually at my disposal'. ,,26

2.1.4 Summary of how opinions are regarded in the Nikayas The above passages that have been selected from the vast corpus of Buddhist Agama literature show a certain consistency of attitude towards the holding of opinions. In much the same way that Gotama Buddha constantly pointed out the hazards of becoming enthralled by material possessions, by sensual pleasures, and by social commitments, he also repeatedly pointed out the hazards of becoming a slave to one's own opinions. In much the same way that he encouraged his followers to reduce their material needs to a bare minimum so that they could travel light, he also encouraged them to pare down their beliefs to a bare minimum. Shaving the scalp in order to eliminate the necessity of carrying combs, shampoos, hair lotions and other grooming supplies,27 the monks also applied Gotama's razor to the inside of the head to clear away all unnecessary thinking and curiosity. Encouraged to meditate on the repulsiveness of food and the physical body in order to curb the appetite for food and sex, the followers of the Buddha were also encouraged to contemplate the disadvantages of holding opinions and forming judgements about things. Seeing that holding opinions leads one to disdain those who hold differing opinions, the monk should try to see the advantages of giving up the habit of passing judgements and having opinions and formulating theories, all of which serve as shac,kles binding

Rational Skepticism in Pre-Dirinagan Buddhism

51

one to the world of turmoil and preventing one from experiencing inner peace. The position that has just been outlined need not yet be regarded as an extreme form of skepticism, such as the type that eventually developed in the Hellenistic world and is apparent in the statements of such philosophers as the Skeptic Metrodorus of Chios, who is reported to have said "We know nothing, nor do we even know the very fact that we know nothing.,,2s But the outlook of the Buddha does show a resemblance to some features of the less methodical form of skepticism that is attributed to the putative founder of the skeptical school, Pyrrho of Elis. Pyrrho, incidentally, is reported by Diogenes Laertius to have travelled with Alexander the Great to India and to have met with and been deeply impressed by the "gyrnnosophists," who are reported to have lived a life of simplicity in monastic compounds and impressed some of the Greeks and Macedonians as perfect embodiments of the ideals of the philosophy of the Cynics. Diogenes Laertius offers this account of the encounter: [Pyrrho] even got together with the Indian Gymnosophists and with the Magi. This led him to adopt a most noble philosophy, to quote Ascanius of Abdera, by introducing akataiepsia [that is, the inconceivability and unrepresentability of things] and epoche [that is, the suspension of judgment].29

In much the same way that the Buddha Gotama ridiculed the brahmans of

his day for failing to live exemplary lives and thus having an empty philosophy, Pyrrho is supposed to have made light of the Academic philosophers and their fondness for abstractions; for him the measure of a philosopher's merit was the degree to which he lived in accordance with his teachings, and the value of the teachings was measured by whether or not they were capable of serving as a guideline for concrete behaviour in the practical world. Aristocles the Paripatetic offers this account of Pyrrho's philosophy: [Pyrrho] did not leave behind any writings, but his disciple Timon says that those who wish to be happy ought to consider three things: in the first place, what is the nature of things; in the second place, in what way must we act towards them; in the third place, what are the results to someone who acts in this way. 1) Well, he says that Pyrrho shows that things are equally indifferent, incommensurable. and indiscriminable, and because of this neither our sensations nor our opinions can be true or false. 2) Consequently, there is no need 10 reconcile or trust them but we must be without opinions. without inclinations. without agitation, by affmning of each thing that it is no more than it is not, or that it is and it is not, or that it neither is nor is not. 3) Those who are placed in this disposition will reach, says Timon, fIrst aphasia and then ataraxia.3D

52

Chapter Two

This aphasia, or non-assertion, that Pyrrho recommends is described by his later systematizer Sextus Empiricus as "the avoidance of assertion in the general sense in which it is said to include both affirmation and negation, so that non-assertion is a mental condition of ours because of which we refuse to affirm or deny anything. ,,31 And the ataraxia that Pyrrho and his followers sought was a state of inner quiet that resulted from abandoning the habit of forming opinions and judgements. The manner in which abandoning views was supposed to lead to inner peace is described by Myles Burnyeat as follows: The assumptions at work here are reminiscent of Socrates, as is much else inBellenistic moral psychology. The emotions depend upon belief, especially beliefs about what is good and bad. Remove belief and the emotions will disappear; as fear, for example, fades when one is dissuaded of one's belief that the thing one was afraid of is dangerous. At least, to the extent that emotions derive from reason and thought, they must disappear when judgment is suspended on every question of fact and value. This will not eliminate bodily disturbances such as hunger and thirst, nor the tendencies to action which result from the endowments of nature and from upbringing in human society. For they do not depend on reason and thought. But they will be less disturbing without the added element of belief about good and bad, truth and . falsity.32

The parallel of Pyrrhonian ataraxia to the Buddhist view of nirviil}a, which becomes possible only with the abandoning of opinion (dN{i; dirrhi), is striking. (See also p. 84 below.) At least in their advocacy of doxastic rninimalism, that is, keeping one's opinions as few as possible, Pyrrhonism and Buddhism appear to have enough of a characteristic in common to warrant our calling one early strain within Canonical Buddhism an inchoate skepticism. Just as Pyrrhonism was given its full articulation by Sextus Empiricus, Buddhist skepticism was given its fIrst apology as a fully developed philosophical system by Nagarjuna, to whom we shall now turn.

2.2 The influence of Nagarjuna Perhaps the most obvious champion of non-assertive skepticism within the fold of early academic Buddhism was Nagfujuna, who devoted a brief treatise entitled "Resolving the dispute" (Vigrahavyiivartanf) explicitly to defending from anticipated criticisms his skeptical position, which is articulated more fully in his major work called MUla-

Rational Skepticism in Pre-Dirinagan Buddhism

53

madhyamakakiirikii. In the opening verse of the Miilamadhyamakakiirikii Nagarjuna pays homage to the Buddha by saying: I pay homage to the finest of speakers, who being fully awakened showed happiness as not coming to an end, not coming into being, not being cut off, not being everlasting, not a single thing, not many things, not approaching, not receding, but as dependent origination, the quelling of vain thinking.33

That same work ends with this dedication to the Buddha: I prostrate before Gautama, who, after experiencing compassion, pointed out the true wisdom in order to dispel all opinions.34

2.2.1 Miilamadhyamakakiirikii Let us look first of all at the opening verse that was cited above. In it we are told that the Buddha pointed out that true happiness (siva) is to be found in quelling vain thinking (prapancopasama). And in the last verse we are told that the aim of the Buddha's teachings was to dispel all opinions (dr~ti). As we shall see below (3.2), when Nagarjuna claims that the greatest happiness consists in abandoning all opinions, there are particular kinds of opinions that he has in mind, namely, those that presuppose the existence of an abiding self (iitman; attan) or substance. Stating the matter in just that way, of course, invites one to reflect upon whether it is possible to form opinions or indeed to think at all without presupposing the existence of an abiding self or substance. It can be argued that it is in the very nature of thinking that the content or subject matter of thought is something that is assumed to be enduring. That this is in the very nature of thinking is a matter about which there is very little controversy in the Indian tradition, including among the Buddhists. Where controversy arises is over the question of whether or not that assumption is true. Is it, in other words, the case that there really are, in the world as it exists independent of our thinking about it, things that endure for us to think about? The general Buddhist answer is negative, as we shall see in greater detail in later sections. But even setting that whole issue aside for the time being, it is clear that for Nagfujuna the key to true happiness lies in the abandoning of all vain thinking, whatever "vain thinking" might be. 35 This is made clear in such statements as the one in Miilamadhyamakakiirikii 18:5: Liberation is due to the removal of afflictions (kle.fa) in the form of deliberate actions; afflictions in the form of deliberate actions are a

54

Chapter Two result of pondering; they are a result of vain thinking. But vain thinking is eradicated in emptiness. 36

Given that "vain thinking is eradicated in emptiness," the issue that naturally arises next is that of determining what this emptiness is that serves to eliminate vanity in thinking. An answer to this question is given in Mulamadhyamakakarika 24:18, which says "We claim that dependent origination is emptiness. To be empty is to be a derivative idea. That alone is the middle path [that avoids extremes].,,37 What it means to say of something that it is a derivative idea is a question that we shall pursue more fully in the next chapter dealing with Buddhist nominalism, but for the present it is sufficient to say that an idea is derivative when it is capable of being reduced to more simple or primitive ideas. Speaking very roughly, we can say that primitive ideas are ideas of primitive things, and derivative ideas are ideas of complex things. Many ideas that we take to be primitive, such as the notion of a self or of identity in general, can be shown, according to most Buddhist thinkers, to be derived from more basic ideas and therefore are not primitive at all. To show that an idea is derivative is approximately the same as showing that the thing to which an idea corresponds is composed of simpler parts and thus is dependent upon those more basic parts for its existence as a complex object. To show that something is composed of simpler parts and is thus dependent upon those parts is to show that it is empty. To say of something that it is empty is to say of it that it is devoid of a nature that is exclusively its own. Insofar as its nature is dependent upon things other than itself and it has no nature that is exclusively its own, it is empty. Realizing that a thing is empty in this sense is believed to be able to help one to break free of the emotional effects, such as fear or desire, that objects often produce when grasped by an unreflective mind. When, for example, a monk is bothered by lust towards a woman, he is invited to reflect upon the fact that since she is made up of physical parts, her overall attractiveness is a product, or at least function, of the attractiveness of her parts. Her face would cease to be lovely, for example, if she were to lose several teeth or her nose or an eye or an ear, or if her hair were to fall out, or if she were to acquire a disfiguring scar. Her body is likewise made up of parts that considered individually fail to inspire much excitement, for it is nothing more than fat and muscle and grease and various internal organs stuffed into a bag of skin that is itself covered with little hairs and out of which grow hard finger- and toenails. This hairy bag full of flesh and fat gives off offensive aromas and dribbles all manner of unpleasant secretions from its pores and openings. Taking all this into consideration, it is clearly a mistake to think that the woman as a whole is attractive, since she is made up of parts that would be regarded as quite disgusting if, for example, they were to fall into

Rational Skepticism in Pre-Dinnagan Buddhism

55

one's food. The woman's beauty, then, is empty, and realizing that it is empty can help the monk not be affected by it. The emotional effects invoked by objects examined uncritically can be eradicated by becoming habituated to examining them critically and analytically. That this is so in the case of physical objects is fairly obvious. But it is no less true of ideas themselves that they can be disassembled into simpler units and that these units can be shown to be untenable or uncompelling. Let us suppose that a system of philosophy can be dismantled into its constituent theories, and each of the theories can be reduced to a number of fundamental assumptions, and the assumptions can all be shown to be somehow inadequate and unworthy of taking seriously; if that can be done, the system of philosophy itself loses most of its appeal and is likely to cease to have much influence on intelligent people. What most Indian systems of philosophy attempt to do is to show that the way of seeing the world that ordinary unreflective people have is inadequate in some way. Insofar as inadequate ways of seeing the world can be replaced by better ways of seeing, people can be given an improved chance at being happy, since unhappiness is universally in Indian thought regarded as a product of faulty understanding. But what most systems of Indian philosophy have in common is an assumption that there are reliable criteria according to which we can determine which of two ideas or theories or, systems is preferable. It is this assumption that Nagfujuna challenges in various ways. Accordingly, not only is the ordinary person's unreflective way of seeing shown to be inadequate, but so is every philosophically more sophisticated substitute. Right understanding, then, consists in realizing that there is no understanding that is less immune from being successfully dismantled than common sense understanding itself, and so there is really no understanding at all that is capable of leading us from painful existence into nirv(ilJa. Consequently, says Nagarjuna, we should realize that the teaching of·the Buddha is just this: True happiness (siva) consists in quelling vain thinking, which consists in quelling all apprehension. Nowhere was any property of anything pointed out by the Buddha. 38

It seems, then, that at the heart of Nagarjuna's programme for attaining nirvalJa is a method of systematically divesting oneself of all beliefs. In order to get some small taste of the style of reasoning that Nagarjuna employs on the way to showing the way to discarding all beliefs, let us look at his treatment of the relation between substance and defIning attribute. This· topic in itself is of no great importance for Nagfujuna, but his brief treatment of the issue is perfectly typical of his approach to every philosophical question. By analysing this approach we

56

Chapter Two

shall be in abetter position to see how he strikes at the foundations of all understanding. In giving an account of Nagarjuna's philosOphical method, I must first state a few obvious points. Whenever we claim to have some understanding of the world, that understanding can only be in the form of a proposition. A proposition is the recognition that a given thing either has or does not have some given property. And every recognition of this type presupposes that we have already determined the precise nature of the given thing and given property that the proposition is about. To take a specific example, suppose that we come to the understanding that space is infinite, or in other words, that space has the property of infinity or limitlessness. In order for this judgement to have any meaning at all, it must be determined what precisely space is and what exactly a limit is. And our determination of what exactly space is involves our recognizing that there is some property that space and space alone possesses; this property, if we succeed in isolating it, is what we may then call the defining attribute of space. To understand anything at all about space requires that we flIst know what space is, and to know what space is requires that we know its defining attribute. So it may be said in general that all understanding is based upon the fundamental supposition that there are substances and that each of these substances has a defining attribute. What Nagarjuna sets out to investigate is the relation between a substance and its defining attribute. What I intend to do here is simply to outline Nagarjuna's argument, rather than to make any assessment of its soundness. 39

Space, according to the standard account of it in Nagarjuna's time, is that element whose defining attribute is penetrability. In other words, it is the element through which solid objects can move without obstruction. When the issue is stated in this way, it sounds very much as if we are saying that space is one thing and penetrability is another. More particularly, it sounds as if space is one type of thing, namely, an absolutely primitive substance, or in other words an element (bhuta), while penetrability is a different type of thing, namely, a property or an attribute. And to say that the substance is one thing while the attribute is another is to say that the substance and the attribute are independent of one another, that is, that each has its own distinct existence without reference to the existence of the other. And if two things are independent of one another, then it is possible to think of one of the things being prior to the other. So let us ask whether space can exist prior to the defining attribute of space. Assume it can. We may now ask whether space exists as space or as nonspace before it acquires the defining attribute of space. Assume that it

Rational Skepticism in Pre-Diimagan Buddhism

57

exists as space. This means that space does not require the defining attribute at all to exist as space; and if we allow this, then we might as well allow that any substance can exist as anything without having the defining attribute of that thing. A rock, for example, could exist as a lluman being without having the attribute of humanity. But this is obviously unacceptable. So let us abandon the assumption that space is space before it acquires the attribute that defines it as space. Let us assume instead that space is not space before it acquires the attribute of penetrability. But if we say that non-space acquires and then bears the defining attribute of space, we are saying in effect that non-space is space, for to bear the defining attribute of space is to be space. But if we allow that non-space is space, we might as well again allow that a rock is a human being. So we must abandon also the assumption that space exists as non-space prior to acquiring penetrability. But if space is neither space nor non-space, then it is nothing at all prior to acquiring penetrability. And if it is nothing at all, then how can it possibly acquire the attribute of penetrability or. any other attribute? The conclusion of all this is that it makes no sense at all to say that the substance space exists prior to its defming attribute. Having argued to this conclusion, Nagarjuna adduces a parallel set of arguments to arrive at the conclusion that the defining attribute is equally incapable of existing prior to the defined substance space. But if neither space nor penetrability can exist prior to the other, we cannot say that they are independent of one another. The substance and its attribute cannot in the final analysis be distinguished from one another after all. And so, the intuition with which we began, namely, that space is an elemental substance that is related to a property with an existence independent of the existence of space itself, proves not to be a reliable intuition after all. In the final analysis, the most fundamental buildingblocks of understanding, the notions of existence and non-existence, must be given up as unsupportable. But giving them up means giving up understanding itself. Not to give up understanding is not to experience the happiness of nirviif)a. So Nagfujuna sums up his argument on the elements by saying "But they of little insight who behold the existence and nonexistence of things do not behold true happiness, which is the quelling of what is capable of being beheld. ,,40 Nagarjuna's treatment of the supposed relation between substance and attribute is quite typical of his treatment of philosophical issues. The first step in his treatment of a problem is nearly always to begin with an apparently satisfactory distinction between two things, such as substance and essential property, substance and accidental property, essential and accidental properties, cause and effect, virtue and vice, happiness and

58

Chapter Two

misery, or ignorance and wisdom. The next step is to demonstrate that these apparently satisfactory distinctions cannot be maintained under close examination, and this for one of two reasons. Either making the distinction ultimately entails a logical contradiction, or the problem we intended to solve by making the distinction in question is in fact not solved, so the distinction turns out to be of no theoretical value to us. As we watch Nagarjuna demonstrate the ultimate untenability of one intuitively satisfactory distinction after another, the message slowly begins to take shape in the reader's mind that the most obvious of "truths" tum out not to be true at all. As we saw in chapter one, Stcherbatsky saw in the Madhyamaka philosophy of Nagarjuna an "extreme skepticism" and mysticism, Claiming that for Nagfujuna "the only source of true knowledge is the mystic intuition of the Saint and the revelation of the new Buddhist Scriptures, in which the monistic view of the universe is the unique subject. "41 According to Stcherbatsky, Nagfujuna's "merciless condemnation of all logic" was done in order to make one receptive to the higher truths conveyed in the new revelations of the Mahayana siitras. But as we shall see below (p. 61) when we consider the Vigrahavyavartanf, there is nothing mystical about Nagarjuna's assault on conceptual understanding. That is, he was not demonstrating to us the frailty of ordinary modes of thinking for the purpose of awakening our minds to the possibility of extraordinary modes of understanding; he was not trying to show that rational thought is limited and inferior to some other type of direct insight that is free of these limitations. On the contrary, Nagarjuna is repeatedly committed to the position that no putative insight into anything can claim to be accurate, true or meaningful unless that insight is logically consistent. No hypothesis that directly or through entailment results in a violation of the laws of contradiction and excluded middle can possibly be a correct hypothesis. And so, if we fail to come up with logically consistent insights into the nature of things, we have no choice but to admit that in the final analysis we do not really understand anything at all. To know or to understand requires that we apply concepts to our experiences and that we manipulate those concepts against one another. And the formation of concepts requires that we make distinctions; a concept as soon as it is formed immediately divides the universe into two parts, namely, the part to which the concept applies and the part to which the concept does not apply. A concept that does not make such a division is meaningless. But if we take Nagfujuna's arguments seriously, they lead us to the conclusion that the most obviously useful conceptllal categories cannot stand up to the test of reason. And what this failure of conceptual

Rational Skepticism in Pre-Diimagan Buddhism

59

distinctions indicates is that the real world does not have divisions in it that correspond to the divisions within our understanding. The real world does not have distinctions of substance and attribute, cause and effect, actor and action, knowledge and thing known, knowledge and ignorance, virtue and vice, suffering and the release from suffering. Thus NagiiIjuna argues: The world of distress (samsara) has nothing whatsoever that distinguishes it from the cessation (nirvalJa) of distress. The cessation of distress has nothing whatsoever that distinguishes it from the world of distress. The boundary of the cessation of distress is the very same boundary as that of the world of distress. Between the two there is nothing whatsoever, not even something of the greatest subtlety.42

And so to understand the world at all is, for Nagarjuna, to be ignorant, or in other words to persist in trying to see the world in ways that the world cannot actually be. In order to gain freedom from dissatisfaction, we must abandon the belief that we can understand the world-as-experienced. 2.2.2 Vigrahavyavartanf

As we saw above, when Nagarjuna says that something is empty he means that it is dependent upon things other than itself for its existence and thus has no identity that belongs to it exclusively. But the word "emptiness (sunyatii)" also has connotations of desolation, worthlessness, and even non-existence. And to say of a statement that it is empty is to say that the statement is false or meaningless. This leads to a problem that NagiiIjuna feels he must address. The problem is that he has gone on record as saying that all things are empty. This invites one to wonder whether the statement that all things are empty is itself a thing. If so, if the statement is true, then it must be an empty thing, in which case it would be either meaningless or false. But if it is not an empty thing, then it is a counterexample to the very claim that it makes, and so the claim that it makes must be regarded as false. So it would appear that no matter how one looks at it, the statement that all things are empty is false or meaningless. Nagarjuna lays this argument out in his Vigrahavyiivartani, along with several other arguments that one might put forth against the notion of universal emptiness. In replying to the anticipated objections to his claim of universal emptiness, NagiiIjuna says this: If I had any hypothesis, then I would have this flaw [of self-

contradiction]. But I have no hypothesis. Therefore I do not have this flaw at all. If I apprehended anything, I might make an assertion or a denial by means of objects that are sensed etc. [seil. reasoned out,

60

Chapter Two learned through analogy or revealed through scriptures]. Since these [objects] do not occur, I have no apprehension of them.43

After stating that he has no hypothesis (pratijfiii), which should be understood as meaning that he has nothing that he wishes to prove, nothing for which he desires to adduce any evidence one way or the other, he then undertakes to examine the whole question of how it is that one can go about deciding that a given belief is true. The general answer to the question of how it is that one decides whether a given opinion is true is that one "measures" it against a standard or a criterion. 44 But the question that Nagfujuna invites us to ponder is that of how the criteria themselves are chosen to serve as a basis of making further decisions. And suppose your confirmation (prasiddhi) of these various conclusions is by means of criteria (pramiil'Jatas); now tell me how these criteria are confirmed. If the confirmation of the criteria were through further criteria, then there would be no end [to the search for the ultimate criterion]. In that case there is no confmnation of the beginning, the middle or the end. If their confirmation is without criteria, then one gives up rational discussion (viida). There is a discrepancy [in that some things require criteria to confirm them, while other things, namely the criteria themselves, need no criteria to establish them].j and a special reason should be stated [to account for this discrepancy]. 5

After giving this argument NagaIjuna anticipates a view whereby the criteria may be self-evident in much the same way that fire is selfilluminating. In just the same way that fire illuminates both itself and those objects around it, says the imagined opponent, a criterion might rationally confIrm both itself and those things that are measured against it. To this suggestion, Nagfujuna offers the observation that there is no better reason for saying that fire illuminates other things by dispelling darkness than that darkness obscures other things by dispelling light; similarly, there is no better reason for saying that a criterion removes doubt from an opinion than that an opinion might cast doubt upon the criterion. 46 Moreover, say:; Nagarjuna, we might come to have confIdence in the criteria because it is through them that we arrive at judgements that seems satisfactory to us. But what this means is that the conclusions we reach are justifying the criteria and not the other way round. If the criteria are to be regarded as sound, their soundness must be established quite independently of any putative knowledge that arises by applying them. The conclusion of all these considerations is expressed by NagiiIjuna as follows: The criteria are not at all confmned by themselves, nor mutually, nor through other criteria. nor are they confmned by the bits 'of knowledge

Rational Skepticism in Pre-Diimagan Buddhism [supposedly ,Justified accidentally.4

through

them],

nor are they

61

confIrmed

In his own prose commentary to these verses Nagfujuna spells out in full exactly what he means in this verse. Supposing that one has an opinion and then is asked how that opinion can be confirmed as true. The person might reply "I know that my opinion that P is true, because I experienced P directly through my own senses." But, says Nagarjuna, we might now ask how it is that we know that sensation itself is reliable enough to confirm anything. That sensation (pratyak$a) is reliable is not, he says, confirmed by sensation itself, nor is it confirmed by reasoning (anumiina), nor is it confirmed by analogy (upamiina) nor is it confirmed by scriptural tradition (iigama) , nor is it confIrmed by a combination of sensation, reasoning, analogy and scriptural tradition. Nor can we justify our opinion that sensation is reliable on the basis of its being the source of what we take to be true, since that would be reasoning in a circle and assuming the truth of the very matter about which there is doubt. Similarly, if a person were to say "I know that my opinion that P is true, because I reasoned Pout through sound argumentation," we might ask how it is that we know that sound reasoning is reliable. Its reliability is not established through sensation, through reason, through analogy or through scriptural tradition. The same argument is given mutatis mutandis to question the reliability of analogy and scriptural tradition. It is, incidentally, because Nagfujuna has no more confidence in scriptural tradition than in anything else that we can dismiss Stcherbatsky's claim (see above p. 58) that Nagfujuna's "merciless condemnation" of logic was done to make the members of his Madhyamaka school receptive to a special mystic intuition or some kind of .revelation. The most natural conclusion to be derived from these passages in Nagarjuna's works is that for him there are no ultimately reliable criteria on the basis of· which we can arrive at sound judgements as to which opinions are worth having and which are not. For this reason, all opinions are to be abandoned by the process of subjecting them to close scrutiny and discerning their ultimate ungroundedness or emptiness (sunyatii). What this implies is a state of suspending all judgement indefinitely, remaining open-minded indefInitely, and never seizing upon any theoretical understanding or opinion as an adequate approximation of the way things really are. Above all, one should be careful not to use the very means by which one divests oneself of all opinions as the basis of a new dogmatism, a dogmatism founded upon an attachment to the view of universal emptiness. For as Nagarjuna says in the Mulamadhyamakakiirikii,

62

Chapter Two The victorious [buddhas] have declared emptiness the antidote to all opinions. But they call incurable those who make an opinion of emptiness. 48

On the other hand, says Nagarjuna at the end of the Vigrahavyiivartanf, those who take this antidote in the proper way can then achieve an goals (sarviirthii/:t), whether they be the goals of ordinary people (laukika) or the goals of the philosopher, which are loftier than those of ordinary people (lokottara). For the person who takes the antidote of emptiness properly the goals of the ordinary person, which involve distinguishing the good life from the bad (sugatidurgativyavasthii), are accomplished. Although these matters can only be fully understood by each person individually (svayam adhigantavyii/:t), some partial understanding can be gained by listening to someone explain things. And this, claims Nagiirjuna in the MiUamadhyamakakiirikii, is the only reason that the Buddha overcame his initial reluctance to teach, a reluctance founded upon the realization that dullwitted people would misunderstand his teaching and potentially do harm to themselves as a result of the misunderstanding by forming unnecessary dogmas and opinions based upon his own words: Emptiness misunderstood destroys the slow-witted like a snake that is mishandled or a craft that is misapplied. And therefore, after reflecting the slow, the sage's thought on the Dharma's being misunderstood turned away from teaching the Dharma. 4

'V

The Buddha, in Nagarjuna's portrayal of his teachings, was willing to take the risk of being misunderstood only because he thought that by teaching he might be able to help a few other intelligent people, whose minds were somehow receptive to his teachings, cure the disease of having opinions. I hope by now to have made a reasonable case for holding not only that this was Nagiirjuna's view of the purpose of the Buddha's teaching but also that such a view of the Buddha's purpose in teaching is consistent with one of the portraits of Gautama ~uddha that emerges in the Nikaya literature. It is this aspect of Buddhist teaching that I think can accurately be called a form of skeptical rationalism.

Notes

63

-- Notes--

Chapter 2. Rational Skepticism in Pre-Dh'magan Buddhism 1. The PaIi term "yoniso" (Sanskrit yoniSas) is an adverbial form of the noun "yoni," which means womb or source. Similarly, the English "principle" derives from the Latin "principiwn," which means foundation, groundwork or primary elements. Thus "principled" is a more literal and in many respects a more meaningful translation of "yoniso" than the ones often given by some of the scholars from the Pali Text Society who have used "orderly," "wisely," and "properly." I owe it to my colleague L.C.D.C. Priestley for pointing out the close correspondence between the Latin and Pali terms and suggesting the felicitious translation of this Buddhist technical term.

2. In the notes that follow. the text of the Sutta Nipata as estabIishec1 by Andersen and Smith (1913 ed., pp. 156-158) is given. In order to facilitate the comparison of my translation with those of others, the Pali is followed by the translation of FausbOll (1881 tr., p. 148-149) and the translation of Norman (1984 tr., p. 135). COIhparing the different translations will reveal how rich in suggestion, and how difficult of precise interpretation, these verses are. 'Paraman' ti dittmsu paribbasano yad uttariIhkurute jantu loke, "mna" ti anne tato sabba-m-aha: tasma vivadani avItivatto. FausbOll (1881 tr.): "What one person, abiding by the (Philosophical) opinions, saying, 'This is the most excellent,' considers the highest in the world, everything different from that he says is wretched, therefore he has not overcome dispute." Norman (1984 tr.): "When, abiding in his (own) opinions, (thinking) 'It is the highest', a person esteems it as the best in the world, then he says all others are inferior. Therefore he has not passed beyond disputes." 3. Andersen and Smith (1913 ed.): Yad attanI passati anisaIilsaIi1 dit(he sute sIlavate muty va, tad eva so tattha samuggahaya nihInato passati sabbam aflnaIi1. FausbOll (1881 tr.): "Because he sees in himself a good result, with regard to what has been see (or) heard, virtue and (holy) works, or what has been thought, therefore, having embraced that, he looks upon everything else as bad." Norman (1984 tr.): "Whatever advantage he sees for himself in what is seen and heard, in virtuous conduct and vows, or in what is thought, grasping at that very thing there, he sees all the rest as inferior. "

64

Chapter Two

4. Andersen and Smith (1913 ed.): Tam vapi gantham kusala vadanti, yam nissito passati mnarn aiiiiam, tasma hi ditthim va sutaIil mutam va sI1abbatam bhikkhu na nissayeyya. Fausbtill (1881 tr.): "The expert call just that a tie dependent upon which one looks upon anything else as bad. Therefore let a Bhikkhu not depend upon what is seen, heard, or thought, or upon virtue and (holy) works." Norman (1984 tr.): "That very (view) the experts call a tie, dependent upon which he sees the rest as inferior. Therefore a bhikkhu would not depend upon (anything) seen, heard or thought, or virtuous conduct and vows." 5. Andersen and Smith (1913 ed.): Ditthim pi lokasmim na kappayeyya iial}ena va sI1avatena va pi. 'sarno' ti attanam anupaneyya 'hIno' na maiiiietha 'visesi' va pi. Fausbtill (1881 tr.): "Let him not form any (Philosophical) view in this world, either by knowledge or by virtue and (holy) works, let him not represent himself equal (to others), nor think himself either low or distinguished." Norman (1984 tr.): "Nor would he form a view in the world because of knowledgt?, or virtuous conduct and vows. He would not represent himself as equal, nor would he tj:Iink of himself as inferior, nor as superior." 6. Andersen and Smith (1913 ed.): AttaIil pahaya anupadiyano illil}e pi so nissayam no karoti, sa ve viyattesu na vaggasan, ditthim pi so na pacceti kiiici. Fausbtill (1881 tr.): "Having left what has been grasped, not seizing upon anything he does not depend even on knowledge. He does not associate with those that are taken up by different things, he does not return to any (philosophical) view." Norman (1984 tr.): "Abandoning what has been taken up, and not taking it up (again), he would not depend even upon knowledge. He indeed does not follow any faction among those who hold different opinions. He does not believe any view at all." 7. Andersen and Smith (1913 ed.): yassubhayante pal}idhidha n' atthi bhavabhavaya idha va huram va nivesana tassa na santi keci dhammesu niccheyya samuggahltll.

Notes

65

FausbOll (1881 te.): "For whom there is here no desire for both ends, for reiterated existence here or in another world, for him there are no resting-places (of the mind) embraced after investigation amongst the doctrines (dhammesu)." Norman (1984 te.): "If anyone has made no resolve in respect of both ends here, for the sake of different existences here or in the next world, he has no c1ingings (to opinions) grasped from among the doctrines, after consideration." Homer and Rahula (in Norman 1984 te., p. 135): "For him who has no desire for renewed existence here or beyond, for him there are no abidings whatever that he would take after having decided among opinions." 8. Andersen and Smith (1913 ed.): tassldha ditthe va sute mute va pakappitl1 n' atthi aJ;IU pi sanna: tam brahmaJ;Iam diHhim anildiyana kenldha lokasmim vikappayeyya. FausbOll (1881 te.): "In him there is not the least prejudiced idea with regard to what has been seen, heard, or thought; how could anyone in this world alter such a BrahmaI:Ja who does not adopt any view?" Norman (1984 te.): "By him not even a minute perception has been formed here in respect of what is seen, heard, or thought. How could anyone here in this world have doubts about that brahman, who does not adopt a view." 9. Andersen and Smith (1913 ed.): Na kappayanti na purekkharonti, dhamma pa tesam na paticchitase, na brahmaJ;Io sIlavatena neyyo, paraIilgato na pacceti tl1rn ti. Fausboll (1881 te.): "They do not form (any view), they do not prefer (anything), the Dhammas are not chosen by them, a BrahmaJ;I is not dependent upon virtue and (holy) works; having gone to the other shore, such a one does not return." Norman (1984 tr.): "They do not form (opinions), they do not prefer. Nor do they adhere to doctrines. A brahman is not to be inferred by virtuous conduct or vows. Gone to the far shore, such a one does not come back again." 10. Further passages in the Sutta Nipata that make this point are verses 824-828, 832-834, 837-841, and 879-914.

11. This verse occurs in the Miigandiyasutta of the Section of Octads of the S utta Nipata. The PaH reads:

" '!dam vadami' ti na tassa hoti Magandiya ti Bhagava dhammesu niccheyya samuggahltaIil, passai'\ ca ditthlsu anuggahaya

66

Chapter Two ajjhattasantim pacinam adassaIh."

12. Andersen and Smith (1913 ed.): Safifiavirattassa na santi gantha, pafifiavimuttassa na santi moha: safifiafi ca ditthifi ca ye aggahesum, te ghattayanta vicaranti loke. Fausboll (1881 tr.): "For him who is free from marks there are no ties, to him who is delivered by understanding there are no follies; (but those) who grasped after marks and (Philosophical) opinions, they wander about in the world annoying (people)." Norman (1984 tr.): "There are no ties for one who is devoid of perceptions. There are no illusions for one who is released through wisdom. But those who have grasped perception and view wander in the world, causing offence." 13. Brahmajala sutta, 1:28, Davids and Carpenter 1890 ed., p. 12. Here the Buddha says: "Atthi bhikkave afifi' eva dhamma gambhlra duddasa duranubodha santa p!llJita atakkavacara nipuI).a p!llJ<;Iita-vedanlya, ye Tathagato sayam abhififia sacchikatva pavedeti, yehi Tathagatassa yatha-bhuccaIh vaI).I).aIh samma vadamana vadeyyum. Katame ca pana te bhikkave dhamma gambira... vadeyyum?"

14. Davids and Carpenter 1890 ed., p. 30: "Tayidam, bhikkave, Tathagato pajan1iti: 'Ime ditthitthanii evam-gahita evam-pariimatthii evam-gatika bhavissanti evam-abhisamparaya ti. '"

15. Brahmaja/a sutta 3:72. See Davids 1899 tr., p. 54.

16. See Brahmajala sutta 3:21. Here the opinion that is being put in the mouth of some teachers is that while pleasures of the senses are transitory, abiding in the first state of contemplation is not. The first state of contemplation is that in which the thinker has concentrated the attention on some topic that is investigated analytically and discursively. It is the state of mind of a careful and methodical thinker. 17. Davids and Carpenter1890 ed., p. 28: "Tan ca Tathagato pajanati, tato ca uttaritaram pajanati, tan ca pajananaIh na paramasati, apariimasato c' assa paccattaIh yeva nibbuti vidita, vedananam samudayaii ca atthagamaii ca assadaii ca adInavafi ca nissaraI).afi ca yatha-bhiitaIh viditva anupadii vimutto, bhikkhave, Tathagato. Ime kho te, bhikkhave, dhamma gambhIra duddasii ... " etc., as in n. 12 above. 18. Rahula 1959, p. 2 19. This formula is very frequently recited in the Agama literature. In order to deemphasize the overtones of fulsome praise usually given this passage, I have deliberately translated it in a way much more prosaic than its customary treatment. For those who would like to compare my treatment with a more standard one, the original text as it appears in Morris (1885 ed., p. 180) is given here, along with the rendering of Woodward. The Pali reads: "... so Bhagava arahaIh sanlinasambuddho vijjacarar.lasampanno sugato lokavidii anuttaro purisadammasarathi sattha devamanussanam Buddho Bhagava. So imam lokam sadevakaIh samarakaIh sabrahmakam sassamaI).abrahm!llJim pajam sadevamanussam sayaIh abhiiiiiii sacchikatva pavedeti." Woodward (1960 tr., p. 163)

Notes

67

translates: "'It is he, that Arahant who is a Fully Awakened One, perfect in knowledge and in practice, Wellfarer, world-knower, unsurpassed charioteer of men to be tamed, teacher of devas and of mankind, a Buddha, an Exalted One. He makes known this world together with the world of devas, Maras and Brahmas, together with the host of recluses and brahmins, both of devas and mankind, having himself come to know it thoroughly for himself'." 20. Morris 1885 ed., p. 189: "AJarn hi va Kalama kaIikhitum alam vicikicchitum. Kailkhaniye va pana va thane vicikiccha uppanna. Etha tumhe Kalama rna anussavena rna paramparaya rna itikiraya rna pitakasampadlinena rna takkahetu rna nayahetu rna akaraparivitakkena rna ditthinijjhanakkhantiya rna bhavyarupamya rna sam3l)o no gam ti, yada tumhe Kalama attana va janeyyatha--ime dhamma akusala ime dhamma savajja ime dhamma vififiugarahita ime dhamma samatm samadinna ahimya dukkhaya sarnvattanti ti-atha tumhe Kalama pajaheyyatha." 21. This is Dhammapada 183, often cited as the quintessence of Buddhist discipline: sabbapapassa akaral)3lil kusalassa upasampada sacittapariyodapanam etam buddhana sasanam 22. Tredennick 1954 tr., pp. 7-8. 23. DIgha Nikaya, Tevijja Sulfa 36. Davids and Carpenter 1890 ed., p. 248. 24. Davids and Carpenter 1890 ed., p. 240. 25. Majjhima Nikaya, Bhayabheravasutta. See Trenckner 1887 ed., pp. 22-23. For an English translation, see Homer 1954 tr., pp. 28-29. 26. Trenckner 1887 ed., p. 482: "Ye te Vaccha evam aharnsu: samal)o Gotamo sabbafifiU sabbadassavI, aparisesam iilil)adassanarn pa!ijanati: carato ca me titthato ca suttassa ca jagarassa ca satatam samitarh iial)3dassanarn paccupatthitan ti, na me te vuttavadino, abbhacikkhanti ca pana man te asam abhiitenati." 27. Davids 1890 tr., p. 19, records the answer of the monk Nagasena to King Milinda's question as to why Buddhist monks shave the head. Nagasena says: "A recluse shaves off his hair and beard on the recognition of the sixteen impediments therein to the higher life. And what are those sixteen? The impediments of ornamenting it, and decking it out, of putting oil· upon it, of shampooing it, of placing garlands round it, of using scents and unguents, and myrobalan seeds, and dyes, and ribbons, and combs, of calling in the barber, of unravelling curls, and of the possibility of vermin. When their hair falls off they are grieved and harassed; yea, they lament sometimes, and cry, and beat their breasts, or fall headlong in a swoon--and entangled by these and such impediments men may forget those parts of wisdom or learning which are delicate and subtle." 28. Cited in Reale 1985:312. 29. Cited in Reale 1985:311. 30. Cited in Reale 1985:316.

Chapter Two

68

31. Cited in Reale 1985:324. For a further account of Pyrrho of Elis and how his thought was developed by Sextus Empiricus and other skeptics, see also Hal1ie 1967b and Popkin 1967, p. 450. 32. Bumyeat 1980:132-133. 33. MUlamadhyamakakarika 1:1-2. Nagarjuna 1960 ed.; p. 4:

anirodham anutpadam anucchedam aSliSvatam anekartham ananartham anagamam anirgamam yai) pratItyasamutpadam prapaflcopaSamarh sivam desayam lisa sambuddhas tam vande vadataril varam 34. Mulamadhyamakakarikii 27:30 in Nagarjuna 1960 ed., p. 258: sarvadr~tiprahlilJaya yai) saddharmam adesayat anukampam upadaya tam namasyami gautamam

35. Like the term "drlfli," the term "prapaiica", when used in the context of a Buddhist work is virtually devoid of any precise meaning. Both of these terms may be regarded as variables that are capable of being given a more or less precise meaning by the Buddhist who uses them. Despite being variables, they do have a constant feature, which is that every Buddhist uses these words to connote wrongful uses of the mind. So, whenever we encounter the terms in a given text, all we can know for sure is that they refer to mental habits that have to be got rid of if we are to attain the greatest good. As was stated above (p. 45), for some Buddhist thinkers, wrongful mental habits might consist in holding certain specified views that are contrary to the principal dogmas of institutionalized Buddhism. For others, a wrongful use of the intellect might consist in any sort of analytical thinking. For yet others, it might consist in naive, uncritical thinking. But in the usage of any given thinker, we can never be sure without further investigation just exactly what kinds of mental habits are seen as being impediments to our highest well-being. So in Nagarjuna's verses all we can know for sure, at least before doing a very thorough investigation of his text, is that the terms have undesirable overtones. 36. Nagarjuna 1960 ed., p.149: karmakleSak~yan mo~

karmakleSa vika1patai) te prapancat prapancas tu sunyamyam nirudhyate

This verse is rendered somewhat differently by other translators. Streng (1967, p. 204) offers this: "On account of the destructions of the pains (kle§a) of action there is release; for pains of action exist for him who constructs them. These pains result from phenomenal extension (prapaiica); but this phenomenal extension comes to a stop by emptiness." Sprung (1979 tr., p. 171) renders the verse as follows: "From the wasting away of the afflictions and karmic action there is freedom. The afflictions and karmic action arise from the hypostatizing thought and this from the manifold of named things. Named things come to an end in the absence of being." Kalupahana's translation (1986 tr., p. 266), being considerably more simple, also has the virtue of not forcing a particular philosophical interpretation upon the reader: "On the waning of the defIlements of action, there is release. Defilements of action belong to one who discriminates, and these in tum

Notes

69

result from obsession. Obsession, in its turn, ceases within the context of emptiness." In these different translations one can see some remarkably different interpretations of the term "prapanca" by modern translators. Note that Streng translates it as "phenomenal extension," Sprung as "the manifold of named things" and Kalupahana as "obsession." From these different translations it is apparent that there is not even agreement as to what general kind of thing prapanca is, such as whether it is an attitude or at least some kind of psychological state as Kalupahana's "obsession" suggests, or whether it pertains to a representation of an object or to an object itself considered in its relationship to a cognizing subject. I am inclined to agree with Kalupahana that prapanca is a species of thinking. In the grammatical tradition prapanca is expatiation, and in particular it is a statement that elaborates the full meaning of a tersely stated rule. See Renou 1957:225. I believe that in Nagarjuna's usage it means something somewhat similar to that, namely, the tendency to build elaborated systems of theories; insofar as such elaboration is unnecessary for the task of attaining nirviif}a, since attaining nirviif}a is not a function of merely achieving a correct intellectual understanding of things, the tendency to indulge in theoretical expatiation (prapanca) upon basic opinions (dr~(i) may be seen as a type of vain, unnecessary and perhaps even distracted and thus connterproductive thinking. But my hunch on the meaning of the term "prapanca" as used by Nagarjuna is not based upon enough familiarity with the entire Malamadhyamakakiirikii to be worth much more than just an example of vain thinking itself. 37. Nagarjuna 1960 ed., p. 219: yaQ pratItyasamutpadal) siinyataril taril pracak~mahe sa prajilaptir upadaya pratipat saiva madhyama Streng (1967, p. 213) renders this same verse: "The 'originating dependently' we call 'emptiness'; This apprehension, i.e., taking into account [all other things], is the understanding of the middle way." Sprung (1979, p. 238) renders it: "We interpret the dependent arising of all things as the absence of being in them. Absence of being is a guiding, not a cognitive, notion, presupposing the everyday. It is itself the middle way." Again, Kalupahana's translation is much more straightforward and contains less gratuitous philosophical sophistication: "We state that whatever is dependent arising, that is emptiness. That is dependent upon convention. That itself is the middle path." 38. Malamadhyamakakiirikii 25:24 in Nagarjuna 1960 ed., p.236:

sarvopalambhopasamaQ prapancopaSamaQ sivaQ na kvacit kasyacit kascid dharmo buddhena desitaQ Compare Streng's translation (1967, p. 217): "The cessation of accepting everything [as real] is a salutary (siva) cessation of phenomenal development (prapanca); No dharma has been taught by the Buddha of anything." Sprung's (1979, p. 262); "Ultimate beatitude is the coming to rest of all ways of taking things, the repose of named things; no Truth has been taught by a Buddha for anyone, anywhere." Kalupahana (1986, p. 369) thinks it is wrong to treat the fIrst line of the verse as a separate sentence from the second line. Rather, he thinks the words in the nominative case in the fIrst line should be construed as modifying the word "dharmo" in the second line. Accordingly, he translates: "The Buddha did not teach the appeasement of all objects, the appeasement of obsession, and the auspicious as some thing to some one at some place." The purpose of the passage,

70

Chapter Two

according to Kalupahana, is simply to state that freedom or nirviif}a is not a substantial thing that the Buddha taught to someone. 39. The argument outlined here is essentially the one given in chapter 5 of the Ma/amadhyamakakiirikii. 40. Mu!amadhyamakakiirikii 5:8 in Nagarjuna 1960 ed., p. 54:

astitvam ye tu pasyanti nastitvam calpabuddhayaI:J bhavanam te na pasyanti dra~tavyam sivam 4l. Stcherbatsky 1932, p 10. 42. Mu!amadhyamakakiirikii 25:19-20 in Nagarjuna 1960 ed., p. 234:

na samsarasya nirvliI),at kirhcid asti vise~aQam na nirvaIJasya sarhsarat kirhcid asti vise~aQam nirvliI),asya ca ya kotiI) kotiI) samsaral}asya ca na tayor antaram kirhcit susilksmam api vidyate 43. Vigrahavyiivartanf29-30 in Nagarjuna 1951 ed., p. 127-128:

yadi klicana pratijiia syan me tata e~a me bhaved do~al:l nasti ca mama pratijiia tasman naivasti me dosab. yadi kirhcid upalabheyarh pravartayeyam nivartayeyarh va pratyaksadibhir arthais tadabhavan me 'nupalambhaI:J In his own commentary to these verses, Nagarjuna specifies the four criteria of knowledge (pramiif}a) that are intended by his phrase "sensation and so forth." They are sensation (pratyak~a), reasoning (anumiina), analogy (upamiina) and scriptural tradition (iigama). 44. The English word "criterion" is borrowed directly from the Greek word "kritiirion," signifying a standard that serves as a means of judging, which in tum derives from the verb "krin6," meaning to divide, separate, discern, or decide. Also derived from this same verb are the Greek words from which the English "crisis" and "critical" are derived; a krisis in general was the process of making a dec;ision, and in particular it was a legal trial, whereas a krites was a judge or umpire who had the quality of being kritikos, capable of making a sound judgement. Similarly, the Sanskrit term "pramiif}a" means a standard of measurement and was also used in legal situations to refer to the evidence that each party adduced to support its case. Stephen Toulmin (1958) in the chapters called "The layout of arguments" and "Working logic and idealised logic" makes a case for seeing the canons of proper evidence in a court of law, rather than the idealised systems of logic worked out by logicians, as the natural "logical theory" for most of our everyday reasoning. Toulmin's suggestions, which were not made at all with reference to the philosophical situation in classical India, nevertheless apply very well to the Indian context. In general I think we may get further in our understanding of Indian epistemology if we see it as a kind of informal jurisprudence rather than as a species of formal logic. Certainlyjn the context of NagliIjuna's critique of the pramiil')as, given the similarity in not only basic meaning but also usage between the Greek term "kriterion" and the Sanskrit "pramiif}a" the English

Notes

71

"criterion" makes a good translation of the Sanskrit term in its general usage. In certain contexts, which will be discussed at greater length in subsequent chapters, it may make more sense to translate it as "knowledge" or as "means of acquiring knowledge."

45. Vigrahavyavartanf31-33 in Nagarjuna 1951 ed., p. 128-129: yadi ca pramiiI)atas te teSlirtJ tesam prasiddhir arthanam teSlirtJ punalJ prasiddhiIil briihi katham te pramiiI)anam anyair yadi pramiiI)ail,! pramaI,lasiddhir bhavet tadanavastha ,nadel,! siddhis tatrasti naiva madhyasya nantasya tesam atha pramiiI)air vina prasiddhir vihlyate vadal,! vaiSamikatvam tasmin visesahetus ca vaktavyal,! 46, A similar discussion occurs at MiUamadhyamakakarika 7:8-12,

47. Vigrahavyavartanf 51 in Nagarjuna 1951 ed., p, 138: naiva svatal,! prasiddhir na parasparatal,! parapramaDair va na bhavati na ca prameyair na capy akasmat pramiiI)anam

48. Mulamadhyamakakarika 13:8 in Nagarjuna 1960 ed., p. 108: siinyatll. sarvad.rsunam proktll. nil,!saraI;Iam jinail,! yesam tu siinyatll. drstis tll.n asadhyan babhasire

49. Mulamadhyamakakarika 24:11-12 in Nagarjuna 1960 ed., p, 216: vinasayati durd.r~tli siinyatii mandamedhasam sarpo yatha durgrhIto vidya va dusprasadhitll. alaS ca pratyudavrttam cittam desayitum munel,! dharmam matvasya dharmasya mandair duravagahatll.m

Chapter 3 Nominalism in Pre-Dinnagan Buddhism In taking technical terminology from one philosophical tradition and applying it to ideas that occur in another, there is always some risk of being misunderstood. In order to reduce that risk somewhat it may be advisable to compare how the terminology evolved in the tradition from which it is being borrowed and to point out differences between that tradition and the tradition to which the borrowed terminology is being applied. In the previous chapter an attempt was made to clarify in what sense the term "skepticism" might be applied to the whole enterprise of Buddhist theory and practice. In this section, and in later parts of this work, I shall be borrowing another term from Western philosophy and applying it to the context within which Indian Buddhist thought evolved. The term in question is "nominalism," and it is now incumbent upon me to clarify how this term came to be used in the history of Western thought so that we shall have some basis upon which to compare the family of issues to which the term belongs in the Western tradition with their cousins in the Indian tradition of philosophy. Among the oldest of problems in the Greek philosophical tradition was that of coming to an qnderstanding of that in virtue of which a plurality of particulars come to be regarded as belonging to a single type or class of things. Just one of many examples of this sort of investigation can be seen in the Platonic dialogue called "Euthyphro," in which Socrates and Euthyphro discuss the definition of piety or the holy (hosios).' On being asked what the holy is, Euthyphro first answers by giving an example of a holy deed. Socrates then pushes for a definition rather than an enumeration of instances, saying: Now call to mind that this is not what I asked you, to tell me one or two of the many holy acts, but to tell the essential aspect, by which all holy acts are holy; for you said that all unholy acts were unholy and all holy ones holy by one aspect. 2

72

Chapter Three

73

This "essential aspect" that Socrates strives to get from his partner in dialogue is what in.Greek was called "to eidos, " which literally means that which is seen, such as the form or the figure of a thing, but which more generally means a general form or a type or kind. 3 In sense this Greek word is very close to what in Sanskrit was called "lakSafJa," a term that, like "eidos" was also used in philosophicaJ contexts in the sense of a defining attribute shared by a plurality of concrete particulars in a given class but not possessed by any particulars outside that class. Moreover, "laksm;a" is like "eidos" derived from a verb whose principal meaning is to see or more generally to observe. So in asking Euthyphro to tell him that "essential aspect" in virtue of which all holy things are holy, Socrates is engaged in an investigation very much like that of various Indian thinkers, including the Buddhist specialists in abhidharma, who were concerned to find the defining attributes (laksm;a) of the things that they felt were of importance. Plato continued to make use of the notion conveyed by the word "eidos" in various of his dialogues. In Book ten of the Republic, for example, in his attempt to arrive at the precise definition, that is, the defining attribute or to eidos of imitation, the philosopher begins the inquiry as follows: Shall we, then, start the inquiry at this point by our customary procedure? We are in the habit, I take it, of positing a single idea or form in the case of the various multiplicities to which we give the same name. Do you not understand? I do. In the present case, then, let us take any multiplicity you please; for example, there are many couches and tables. Of course. But these utensils imply, I suppose, only two ideas or forms, one of a couch and one of a table. Yes. And are we not also in the habit of saying that the craftsman who produces either of them fixes his eyes on the idea or form, and so makes in the one case the couches and in the other the tables that we use, and similarly of other things? For surely no craftsman makes the idea itself. How could he? By no means. 4

From this Plato develops his celebrated notion of forms or ideas as separate entities that exist independently of and prior to particulars and are more real than particulars, which are in turn merely imperfect replicas of the archetypical forms; to be fully real is for Plato to be absolutely singular, which is to be indissoluble into smaller parts and hence imperishable. The archetypes have this absolute unity, while ordinary

74

Nominalism in Pre-Dirinagan Buddhism

particulars of sense experience lack it. This consideration leads Plato to posit a division of the world of know abIes into two discrete realms, namely, the realm of phenomena, comprising perishable concrete particulars that are apprehended through the senses, and a realm of imperishable archetypes (arkhai) that are not apprehensible through the senses but are knowable only through the intellect. Moreover, it is only the cognition of these pure forms that constitutes genuine knowledge (episteme or gnosis,) while cognition of the particulars constitutes mere opinion (doxa), especially in a person who has failed to realize that these particulars are onil transitory imitations of the stable and enduring archetypes or ideas. Plato's doctrine of archetypes comes in for some trenchant criticism by Aristotle. Aristotle begins his Metaphysics with a discussion of various types of knowledge, since, as he says in the opening sentence, "All men by nature desire to know. ,,6 The lowest form of knowledge, he says, is sensation (Greek aisthesis; Latin sensus)'? which "makes us know and brings to light many differences between things." Sensation is common to all forms of animal life, but higher forms of animals also have memory. Memory produces experience (empeiria; experientia), which is the unification of sensations into a connected and patterned whole (katholou; universale). In man, says Aristotle, there is a further capacity to form judgements on the basis of reason; judgement unifies many notions that have arisen from experience. This higher synthesis Aristotle calls art (tekhne; ars), which is knowledge of universals, in contrast to experience, which is knowledge of individuals (kath' hekaston; singulare).8 But even art is merely a knowledge of what is the case. It takes a higher kind of knowing, which Aristotle calls wisdom (sophia; sapientia), to understand why things are the case. Wisdom is the knowledge of archetypes or first principles (arkhe; principium) and causes. It is this investigation into causes and first principles that makes up the subject matter of the Metaphysics. Later on in the Metaphysics, Aristotle states that the subject matter par excellence of the science of metaphysics, which is the science that leads to wisdom, is being as such. There is a science which investigates being as being and the attributes which belong to this in virtue of its own nature. Now this is not the same as any of the so-called special sciences; for none of these others treats universally of being as being. They cut off a part of being and investigate the attribute of this part; this is what the mathematical sciences for instance do. Now since we are seeking the first principles and the highest causes, clearly there must be some thing to which these belong in virtue of its own nature. If then those who sought the elements of existing things were seeking these same principles, it is necessary that the .e1ements must be elements of being not by accident but just

Chapter Three

75

because it is being. Therefore it is of being as being that we also must grasp the first causes?

After giving his general account of what the subject matter of metaphysics, the science of wisdom, is, Aristotle reviews the attempts of philosophers before him to come up with some account of first principles (arkhe). Among the previous attempts to arrive at an adequate account of archetY}6es that he considers is that of Plato, whom he criticizes at some length. Plato's doctrine of archetypes, says Aristotle, leads to a proliferation of entities that are not really necessary in order to give an adequate account of the world and our experience of it. If, for example, it is held that there exists an archetype for every particular kind of thing of which we can have knowledge, it would follow that there would be an archetype not only for each type of thing but also for the absence of each type of thing, since we can be aware of absence. Moreover, there arises a great confusion from the fact that in Plato's system of thought the same name applies both to an archetype in the ideal world and to the imperfect replica of it in the sensible world. Not only that, but Plato is accused of failing to give adequate account of the relation between the archetype and its sensible replicas. Are we to suppose, asks Aristotle, that they have the same form in common, and if so are we to suppose that this common form is yet a third thing that exists alongside the archetype and its sensible imitation ? Yet another consideration is that these theoretical archetypes that Plato posits cannot play any role in either the production, that is, the actual being, of particulars or in our acquiring knowledge of particulars. Being supposedly eternal and changeless, the archetypes cannot participate in the movement of particulars nor in their other changes, and being insensible the archetypes cannot be observed along with the particulars in which they are supposed to "participate." Aristotle's criticisms of Plato are not usually understood as constituting an outright denial of archetypes, but rather as constituting a denial of archetypes as substances that exist independently of the concrete individuals in which they occur. He. does say, for example, that "evidently there is a first principle [arkhe; principium], and the causes [aitia; causa] of things are neither an infinite series nor infinitely various in kind."ll But there is, says Aristotle, a problem in granting that universals, which Plato says have being par excellence owing to their being absolutely singular and imperishable, have substance. In discussing Plato's view on this matter, Aristotle points out that if unity and being are not substances, then no universal can be a substance, for unity and being are the most universal of all things; consequently, if unity and being are not substances, then nothing exists but individuals. On the other hand, if unity and being are substances, it is hard to see how there can possibly be anything that exists

76

Nominalism in Pre-Diimagan Buddhism

besides this one universal, namely, being itself. Neither, of these alternatives, says Aristotle, is entirely acceptable. 12 After entertaining a number of other problems involved in these views, Aristotle summarizes the difficulties in the Platonic view of archetypes. We must not only raise these questions about the first principles [arkhe; principium], but also ask whether they are universal [katholou; universale] or what we call individuals [kath' hekaston; singulare]. If they are universal, they will not be substances [ousia; substantia]; for

everything that is common indicates not a "this" but a "such," but substance is a "this." And if we are not allowed to lay it down that a common predicate is a "this" and a single thing, Socrates will be several animals--himself and "man" and "animal," if each of these indicates a "this" and a single thing [henotes; unio]. If, then, the principles are universals, these results follow; if they are not universals but of the nature of individuals, they will not be knowable [episteton; scibile]; for the knowledge of anything is universal. Therefore if there is to be knowledge of the principles there must be other princi~les prior to them, namely those that are universally predicated of them. l

The position that Aristotle is usually said to have taken on this question of universals is that they do not exist independently of the concrete particulars in which they occur, but neither does a concrete particular occur without having some universal aspect about it; every particular must occur in some form (eidos; species) or other, for it cannot occur quite indefinitely, and this form is common to a plurality of individuals and so is universal. If, says Aristotle, the form existed independently of the material (hule; materia)--if, for example, a house existed independently of the bricks, then how could any particular house come into being? The bricldayer makes a particular house with a particular form, and the result is a structure that is particular in one sense and universal in another sense. But the form has nothing to do with actually causing the corning into being of the informed matter, and therefore there is no need to posit, as Plato does, that the forms are independent causal factors that exist prior to the particulars that are patterned after them. Rather, the form in a sense comes into being simultaneously with the particular substance of which it is the form. Aristotle concludes: Obviously, therefore, it is. quite unnecessary to set up a Form as a pattern [paradeigma; exemplum] (for we should have looked for Forms in these cases if in any; for these are substances if anything is so); the begetter is adequate to the making of the product and to the causing of the form in the matter. And when we have the whole, such and such a form in this flesh and in these bones, this is Callias or Socrates; and they are different in virtue of their matter (for that is different), but the same in form; for their form is indivisible [atomon; indivisibi/is].14

Chapter Three

77

Aristotle had numerous interpreters among later Muslim and Christian philosophers. Through the influence of St Thomas Aquinas (1225-1274) especially, Aristotle's logic and metaphysics came to provide a standard framework for the systematic and principled presentation of Christian doctrine. The Thomist interpretation of Aristotle set the tone of what eventually came to be called the old way (via antiqua) of interpretation; in the fourteenth century there arose a "modern way" (via moderna) of presenting Christian theology through the framework of Aristotelian thought. 15 This "modem way" was associated with such thinkers as William of Ockham (d. ca. 1349), who presented a view of universals that has come to be called nominalism, according to which "no universal is a substance existing outside the mind."16 In defending this conclusion, Ockham emp.loyed a number of the same arguments that Aristotle had used against Plato's notion of universals. First of all, says Ockham, a universal cannot be a particular substance, since if it were then an individual such as Socrates would be· a universal. Moreover, says Ockham, every substance is either one thing or many. If it is one thing, it is numerically one. But if it is many things, then it must either be a plurality of universals or a plurality of particulars. Supposing we say that a universal is a substance that is many particulars, it would then follow that a universal could be distinguished from a single particular but not from a plurality of particulars. But if in order to avoid this absurdity we say that a universal is a substance that is a plurality of universals, then we must ask about each one of those universals whether it is one thing or many things. If we say it is one, it would then follow that a universal is a particular, which has already been shown to be an untenable conclusion. But if we say that each of the universals is many things, we again have to ask of each of them whether it is one thing or many, which leads either to an infinite regress or to the absurd conclusion that a universal is a particular. But then, Ockham continues, let us suppose that a universal is a single substance that exists in particulars but is nevertheless distinct from them. If that is the case, he says, then it would follow that there could be universals that exist apart from particulars altogether. And if that were true, then no individual could come into being, for part of it, namely its universal part, would already exist and so could not come into being a second time. Moreover, it would follow that no individual could completely perish, since the part of it that was universal would continue to exist; but if this universal part also perished with an individual, then in perishing it would also destroy all other individuals that had the same universal in them. Ockham corroborates his own arguments with several appeals to the authority of Aristotle and the commentary of Averroes and then concludes:

78

Nominalism in Pre-DiIinagan Buddhism Therefore, it ought to be granted that no universal is a substance regardless of how it is considered. On the contrary, every universal is an intention of the mind which, on the most probable account, is identical with the act of understanding. Thus, it is said that the act of understanding by which I grasp men is a natural sign of men in the same way that weeping is a natural sign of grief. It is a natural sign such that it can stand for men in mental propositions in the same way that a spoken word can stand for things in spoken propositions.!?

All that is required for man to form abstract understanding, says Ockham, is a plurality of individual sensible objects and an intellect; there is no need to posit a further entity such as a form (eidos; species) as existing outside the intellect. The apprehension of a sensible object leaves a trace on the mind; the process by which the natural world leaves these traces on the intellect is unknown to us, but the effect is known to us in the form of subjective abstractions or universals. 18 It is this view that universals do not at all exist outside the mind, even though they may be formed as a result of apprehending things from the external world, that is called nominalism, and it is a view rather similar to this that eventually evolved in the Buddhist world and especially in the philosophy of DiIinaga. It must be borne in mind that the context in which Buddhist nominalism evolved was not quite like the context in which it· evolved in Western philosophy. In classical India there was no position quite like that of Plato, and consequently no positions quite like those that in Greek and Hellenistic philosophy evolved in opposition to Platonism. There are, nevertheless, philosophical positions in India that bear a resemblance in several important respects to positions in European philosophy. In particular, as we shall see later, both the conclusions and even many of the arguments of DiIinaga bear a resemblance to the nominalism of Ockham. While nominalism was not fully developed in early Buddhist thought, there was always a definite tendency towards nominalistic thinking. It is to some of the pre-Dinnaga anticipations of nominalism in Buddhist thought that we shall now tum.

3.1 The Agama literature and Milindapafiha In the Nikaya literature there are two separate issues that eventually serve as the source of a nominalistic theory. One of these issues is that of what constitutes a true brahman, and the other is the issue of whether or not there exists an enduring reality that underlies our" notion of a permanent self (iitman; attan), or an identity that we carry with us

Chapter Three

79

unchanged from conception to death. In arguing that the true brahman is a brahman in virtue of conduct rather than in virtue of family, the Buddha Gotama sets the tone Jor challenging a doctrine of natural classes that was widely accepted in his day; eventually, when this challenge is taken to its logical conclusions, it came to be questioned whether any of our inttiitions about natural classes are based in realities outside the mind. And in arguing that there is no abiding self, he set down another principle that when ,taken to its logical conclusion evolved into the doctrine that nothing that is made up of smaller parts or components is ultimately real, despite its being conceived as real by the mind. First, then, let us see how exactly Gotama Buddha expressed himself in these matters, and then let us see how these ideas developed into a more fully developed doctrine of nominalism. 3.1.1 Natura] class in the Nikayas It is held by some scholars that Gotama Buddha rejected, or at least thoroughly reinterpreted, the Brahmanical concept of caste (jiiti). A.K. Warder, for example, recapitulates the Buddha's attitudes as follows: "All men are born equal, and born equal to the gods, to 'God' Himself. What they become depends on their own conduct. The Brahmanical myths about the origins of the different races and classes of men are non-sense. ,,19 But Y. Krishan (1986) has criticized Warder's views on this point and has argued that neither Jainism nor Buddhism were in fact concerned with social reform or the improvement of social conditions. He cites passages that he thinks indicate that the Buddha condoned the caste system and even untouchability among laymen, but that when people joined the Buddhist Sangha these distinctions of birth were renounced along with all other worldly trappings. Moreover, argues Krishan, by arguing that caste is a function of previous karman, the Buddha unwittingly strengthened the rigidity of the caste system by the doctrine that one's lot in this life was a deserved and inevitable consequence of one's former behaviour. This controversy is not one thilt I wish to enter into here. It should be borne in mind,however, that, as Etienne Lamotte (1958:5-7) has pointed out, the social milieu in which Gotama Buddha taught was probably not that which is described in the mythology of the Upani~ads or the social codes of Manu, and that imputing to the Buddha Gotama views on issues that became of paramount importance many centuries later would be clearly anacllFonistic. Whether this claim is legitimate or not as an expression of the teachings of the Buddha, it is clear that he did make statements that gave later Buddhists a basis for rejecting the caste system.

80

Nominalism in Pre-Diimagan Buddhism

In the Vasetthasutta (Sutta Nipata 3.9), for example, the story is told of two young brahmans who are living and studying in the area of Icchanarhkala, where many prominent and wealthy brahman teachers reside. The two youths, Vasettha and Bharadvaja, get into a quarrel on the issue of whether a person becomes a brahman by nature (jati) or through behaviour (kamma). Bharadvaja takes the stance that in order to be a brahman one must come of parents whose pedigrees are unimpeachable on both sides of the family for at least seven generations back, while Vasettha holds that one becomes a brahman by good conduct (sfla) and vows (vata). Neither being able to convince the other, the two youths approach the Buddha to ask for his views on the matter, and he instructs them as follows: 600. "I shall explain the classification (vibhaiiga) of natures (jali) among living beings to you, Vasettha," said the Lord, "systematically and according to how things are, for natures are mutually distinct. (601) You know grass and trees, and although they are not aware of it, their distinguishing signs (liiiga) are natural (jalimaya), for natures are mutually distinct. (602) Then in the insect and the grasshopper and all the way to the antsd their distinguishing signs are natural, for natures are mutually distinct.2

The Buddha continues enumerating natural divisions within the animal and plant kingdoms, noting that in each case what distinguishes the species one from another are natural features rather than arbitrary divisions. But among human beings there are no such natural distinctions to be found that naturally divide the human race into groups that are distinct from one another. Whereas other members of the animal kingdom are distinct from one another by such distinguishing features as different types of fur, different kinds of eyes and ears, different mouths and types of body, different colours, and different voices, there are no such features that distinguish brahmans from peasants, artisans, merchants, servants, thieves, soldiers, sacrificers or kings.21 A key verse in the midst of this argument is Sutta Nipata 611: "This differentiation (vokara) is not found among embodied human beings as individuals (paccattam). The differentiation among human beings is assigned through names.,,22 It is obvious that in this whole discussion the Buddha is not arguing anything like a nominalist view that says that all distinctions among objects and all classifications of things are merely nominal or conceptual. On the contrary, he is arguing that there are some distinctions that are quite real and are grounded in nature (jati).23 But among these very real distinctions is not the set of distinctions that human beings make among themselves as they divide into social classes, occupational groups, tribes, nations, ethnic groups and all the other arbitrary separations that human beings devise." But in stating things in just this way the Buddha unwittingly set in motion a long

Chapter Three

81

philosophical debate on the vital issue of just which distinctions that we make are legitimate and grounded in nature and which are arbitrary and grounded only in the act of naming (samafiiiii). 3.1.2 Personal identity in the Pali Canon An account of Gotama Buddha's fIrst sermon to fIve monks at Isipatana (Deer Park) occurs in the PilIi Vinaya, Mahavagga 1.6. According to this account the Buddha begins by explaining that his teaching is of a middle path that avoids the two extremes of a life given to the pursuit of pleasures and a life given to self-mortification. He then explains the noble eight-fold path that leads to insight (cakkhu), understanding (fiii/Ja), calm (upasama), superior knowledge (abhififia), full awakening (sambodhi), and nirvaIJa (riibbiina). This leads into a discussion of distress (du/:tkha; dukkha), its cause, its elimination and the means of its elimination. This fIrst sermon results in the five monks becoming the Buddha's fIrst disciples, to whom he then gives further teaching. In this further teaching the fIrst topic covered is that of the nature of the self (atman; attan), about which the Buddha speaks as follows: Monks, the physical body is not the self, because if this physical body were the self, this physical body would not meet with disease, and [the wish] "Let my body be like this. Let my body not be like this." would be granted. And since the physical body is not the self, it does meet with disease, and the wish "Let my body be like this. Let my body not be like this." is not granted. 24

The Buddha then applies this same formula mutatis mutandis to the other four aspects of what we are likely to think of as our selves, namely, our feelings (vedana), perception (samjfia; safifia), character (samskara; samkhiira) and awareness (vijfiana; vififiaIJa). Since each of these things is liable to affliction and change, and since we do not have control over the way that any of these things evolve, none of them can be regarded as the self. The self, then, if there be such a thing at all, would be that which is in control and that which remains constant and unperturbed while things around it change or disappear. If all we knew of the Buddha's teaching on the self (atman; attan) was this one passage, it would not be at all clear what exactly his position was. It could very well be interpreted as implying that while the empirical realm is made up only of things that come into being, change and ultimately perish, and so the self is not to be found in the empirical realm, there is another non-empirical realm in which the objects that become the subject matter of awareness do not perish. The Buddha could, in other

82

Nominalism in Pre-Diilllagan Buddhism

words, have believed in something like Plato's realm of archetypes that was discoverable only by the intellect (see above p. 74), and he could have believed that the self was one of these archetypes. He certainly does not advocate anything like that view explicitly, but there has been some question mooted from time to time as to whether the Buddha might have held such a view in the sense that such a view is presupposed in the remainder of what the Buddha teaches. 25 Given, however, the presence of numerous other suttas in the Pali Canon that deal with this topic of the self, it seems quite unlikely that the Buddha would have endorsed anything like the Platonic view. One of the many suttas in the Nikaya literature that deals with the question of the self is the P ouhapada Sutta of the DTgha Nikaya, where the question is treated in some detail. This sutta opens with an account of a renunciate by the name of Potthaplida, who has been living amidst a group of fellow renunciates who are given to a great deal of loud discussion and debate with one another about various worldly affairs as well as about various speculative matters. Ponhaplida has been impressed upon seeing the Buddha, partly because the Buddha is so fond of silence. So he urges his fellow renunciates to hold their tongues and be quiet so that the Buddha might be willing to keep company with them for a while. When the Buddha does come to visit with them, he senses that they have been busy discussing some point, and he asks what they have been talking about. Ponhaplida, perhaps out of embarrassment at having to admit that what they had really been talking about was politics and other worldly matters, evades that question and poses a somewhat different problem to the Buddha. He says that on a former occasion a discussion had broken out about the nature of catalepsy (abhisannii-nirodha), when all sensation and awareness leaves the physical body in a state of trance. On the occasion of . that discussion various people had had various opinions about the nature of consciousness and how it arises and how it is lost. Some had argued that perception (sarhjnii; sannii) comes and goes without any causes or conditions (ahetu-appaccayii), while others had argued that perception is in fact the self (attan), which sometimes is in the body, making it sentient, and sometimes leaves the body, making it insentient. Still others thought that magicians and others of great power had the ability to render beings sentient or insentient through the application of spells. The Buddha is then asked what his views are on this question. The Buddha's reply to this question of the nature of sentience is that it does not arise without causes and conditions, but on the contrary comes into being and goes out of being as a result of discipline (sikkhii). For example, a person who undertakes self-cultivation through

Chapter Three

83

contemplation (dhyana; jhana) overcomes lust and desire. The sutta goes on to say: "That perception of desire that one had before," said the Lord, "comes to an end. On that occasion there arises a perception, which is both· subtle and real, of pleasure (pIli) and happiness (sukha) that is born of discrimination, and one becomes possessed of a perception that is both subtle and real of pleasure and happiness that is born of discrimination. And so it is that some perceptions arise through discipline and some perceptions perish through discipline. ,,26

The Buddha goes on to describe the other types of contemplative states in similar terms, each time emphasizing that it is a result of discipline that perceptions of one kind are replaced by perceptions of another kind until such time as one grows weary of perceptions altogether. Then one deliberately stops thinking and being conscious altogether and brings about the complete cessation of perception. This leads Ponhapada to ask the Buddha whether he believes there is only one highest state of perception or several highest states. The Buddha's reply is: "By whatever means one reaches cessation [that is, nirva~a], I proclaim that the highest state of perception. Thus I proclaim both that there is one highest state and that there are many highest states of perception. "27 When Ponhapada hears the Buddha's answers to the questions that he has posed so far, he then returns to the question of whether the self is identical with the physical body or whether it is something other than the physical body, that is, something that sometimes dwells in the physical body and leaves it at other times. On being asked this question the Buddha replies with a question of his own: "But what self do you believe in, Ponhapada?"28 Potthapada then outlines a number of alternative views on what the self might be, beginning with the notion that it might be the physical body. The Buddha points out that if the self were the physical body, it would follow that the self and perception are two different things, for one and the same physical body has a plurality of perceptions. It would therefore be the case that the physical body is that which remains constant relative to the coming and going of various states of perception. Ponhapada then suggests that the self might be made up of the mind (manas). The Buddha points out that the same observation he made about the physical body could be made with respect to the mind, namely, that it would have to be different from one's perceptions owing to its remaining relatively constant while perceptions change. This leads Potthapada to conclude that the self must be made up of perception itself. But even in this case, says the Buddha, the perceptions come and go while the self that is putatively made up of perceptions remains stable. Ponhapada thinks all this over and finds it very hard to decide whether he is satisfied with

84

Nominalism in Pre-Diimiigan Buddhism

thinking that the self and perceptions are different things or the same thing. The Buddha simply agrees that this is a difficult thing to understand. Moreover, he says, it is not a very important question; knowing the answer to it, if it has an answer, would not-make a person any happier, nor would it bring a person any closer to nirv(1)a. This answer makes all renunciates who have witnessed this dialogue between the Buddha and Potthapiida think that the Buddha is unworthy of any further consideration. If he cannot answer questions about the real nature of the self, they say, and if he cannot answer questions on how the world was created and other such matters, then he hardly qualifies as a sage at all. Several days after this initial encounter with the Buddha, Potthapiida returns to see him again, this time accompanied by an elephantdriver's son named Citta. The Buddha greets them by saying: "All those renunciates, Potthapiida, are blind and lacking insight (acakkhuka). You are the only one of them with any insight. I teach and make known that . some things are certain, and some things are uncertain."29 That which is certain, says the Buddha, is the nature of distress (dukkha), how it comes into being, its cessation and how to achieve its cessation. That which is uncertain is how the world came into being, whether life is identical with the physical body or different from it, whether one who has realized the truth exists after death or not and questions of that nature. Therefore, on these matters the Buddha has no opinions. He then relates how he went from teacher to teacher and listened to their teachings about the nature of happiness. These teachers spoke of states of happiness that would be realized in life after death. And it became clear to Gotama that the kinds of happiness they talked about were purely theoretical and had never been personally experienced by the teachers themselves. So he rejected their teachings as so much stupid chatter (appatihfrakatam bhtisitam). In contrast to their teachings, he said, his own teaching was one that was well-grounded, for it was based on an experience of happiness that he himself had experienced and that any other person could experience by following the prescribed path of practice. After hearing the Buddha's discussion, Citta the elephant-driver's son states the view that whenever one has an apprehension of oneself (atta-patilabha) as the physical body, one's self as the mind is abandoned (mogha), and whenever one apprehends oneself as made .up of the mind, one's self as the physical body is abandoned. 3o Similarly, the apprehension of oneself in the past was present in the past, and the self that one is now and one's future self was abandoned; when one apprehends oneself in the present, the past self and the future self are abandoned; aI}d when one apprehends oneself in the future, one's past and present self will be

Chapter Three

85

abandoned. After listening to Citta's observations, the Buddha says that it is not quite right to think that what is being apprehended is present and that other things are non-existent, but rather that what one is apprehending is given a name (sankharh gacchati) that does not apply to the things to which one is not attending. When one talks of one's present self, one calls it one's present self, and one does not call it one's past self or one's future self, and one does not call either of these other two selves one's present self. The Buddha offers an analogy from ordinary life?l Milk comes from a cow, he says, and eventually it sours, and from this sour milk one can make butter, and from the butter one can make ghee. One does not call the fresh milk sour milk or butter or ghee, nor does one call the sour milk fresh milk or butter or ghee and so on. These distinctions are merely convenient names. Similarly, when one apprehends one's physical self, one does not call it one's mental self, and when one apprehends one's mental self one does not call it one's physical self. These distinctions are also just convenient names. The Buddha summarizes the importance of their dialogue by saying that each of these names for the self is a "folk name (ioka-samafzfiii), a folk expression (loka-nirutti), a folk designation (lokavohiira), a folk idea (loka-pafzfzatti) by which the Tathagata conducts business without being adversely affected.,,32 Buddhaghosa, incidentally, in his commentary to this sutta takes this summary to mean that the Buddha is capable of using language that really refers to nothing real at all, since the world is in fact empty (sufzfza) and vain (tuccha); but in using these folk expressions the Buddha is not affected by desire (talJhii), conceit (mana) and opinions (dirrhi), whereas ordinary folk are affected by these impediments when they use folk speech.33 There is in this dialogue nothing like. a fully developed philosophical discussion on the ontological s~atus of universals. In fact, there is an explicit statement by the Buddha saying that questions of the general type of which the question of the existence of universals is one instance are unimportant in the quest for nirvalJa. The most that we can expect to find in the Buddha's dialogues, therefore, is that questions are broached on a certain range of topics that will later become the subject matters of more thorough investigations. One particular issue that is broached in the PoUhapada Sutta is that of the relationship between unity and diversity, which is simply acknowledged to be a thorny problem that is so difficult to understand that one should perhaps be wary of teachers who claim to have solved it. But in making the statement that many of the distinctions that are made in our thinking are made on the basis of practical needs, which are reflected in folk language, but that taking these distinctions too seriously might lead one astray if they are used without philosophical caution, the Buddha set in motion the long-standing

86

Nominalism in Pre-Diimagan Buddhism

Buddhist preoccupation with the question of the relationship' between language and reality, a question to which Indian Buddhist thinkers from the time of Nagasena on through to the time of the end of Buddhism in India constantly returned. 3.1.3 Personal identity in the Milindapafiha It was noted above (p. 73) that the Sanslait term "lak$Q1Ja" (Pali "Zakkhar;a") is similar in philosophical usage to the technical term "to eidos," which in the usage of Socrates had the sense of a defining attribute or "essential aspect," that is, the common characteristic by which one could know all the particular members of a class as belonging to that class. In the famous dialogue that supposedly took place in the middle of the second century BeE between the Buddhist monk Nagasena and Milinda, King of Bactria, there is one entire section devoted to Zakkhar;a or defining attributes; in that division the principal topic is the notion of the person (puggaZa) and the person's defining attributes. This dialogue takes place after the King asks Nagasena by what name he is known, and Nagasena replies "I am known, your highness, as Nagasena... and although my parents named me Nagasena or Siirasena or Vlrasena or SIhasena, it is nevertheless a name (sanhka) , an appellation (samafifia), an idea (pafifiatti), a designation (vohara), nothing but a label (namamattam) , for there is no person apprehended here. ,,34 This statement prompts the King to wonder, if there is no person apprehended that corresponds in some way to the proper name, who it is that gives requisites to a monk and who it is that uses those requisites given by donors. He wonders also who it is that is following the monastic code, practising meditation, experiencing the consequences of practice and seeking nirvar;a. And finally he points out that if there is really no such thing as a person corresponding to the name, there would be no harm in killing the monk, for there would be no one there to be killed. But, having said all that, the King then begins a systematic search for the elusive person who might answer to being Nagasena. The person of Nagasena is not be to identified with anyone of his physical parts nor with the totality of his physical parts, nor is he to be identified with feelings (vedana), whether considered one at a time or as a group, nor is he to be identified with perception (sarhjfia; safifia), character (sarhskara; sarhkMra) or awareness (vijfziina; vififiar;a). But while Nagasena is not to be found among any of these phenomena, neither is he to be found anywhere outside them. Nagasena, he concludes, is merely a word (saddo yeva). Therefore, concludes the King, Nagasena must have been lying when he said he was Nagasena.

Chapter Three

87

Nagasena then asks the King how he had conveyed himself from his palace to the place where this discussion was being held. The King replies that he travelled by chariot. Nagasena asks the King questions about his chariot that were parallel to questions that the King had asked him about himself. Is the chariot found in the wheel, or is it found in the chassis, or in the axle, and so forth? None of these things is where the chariot is, says Milinda, and then Nagasena concludes that the King must have been lying when he said he came by chariot. King Milinda was apparently one of that rare breed of political leader who had a sense of hUI11our, for he allows Nagasena to continue the debate and to make his point that the notion of a person is like the notion of a chariot. Nagasena's argument is sketchy, but if we were to spell it out in full it would probably be something as follows. That which we call a person can in fact be analysed into discrete components, just as the chariot can be disassembled. Anyone of these components may be altered or replaced or deleted without impairing the supposed integrity of the collection of those parts. For example, a chariot's wheel can be replaced without altering the chariot's "identity," that is, without making it a different chariot. And of course it is well-known that a person's body undergoes changes all the time, and it is acknowledged that habits can be replaced by others, and knowledge can be gained or lost, and all these changes can occur without changing our sense of who that person is. The person's "identity," in other words, remains intact through all this change. But when we inquire into where this so-called identity resides, we find that it cannot reside in its totality in anyone component part, nor can it reside in the set of parts taken as a whole. For if, let us say, the entire identity of the chariot were to reside in a single bolt, then the chassis and the axle and the wheels would not be parts of the chariot at all, for they are not parts of that one bolt. The chariot would be just the bolt. And if the bolt should break and be replaced, we should have to say that the entire chariot was broken and replaced by an entirely different chariot. On the other hand, if we assume that the identity of the chariot resides in the collection of parts taken as a whole, then, since the whole changes any time any part changes, the replacement of any part would be to change the identity of the whole; and once again we should end up having to accede to the absurd conclusion that to replace a single bolt in the chariot would be to create a wholly different chariot. The notion of identity, then, is one that leads to all sorts of difficulties if we try to extend it beyond the realm of the imprecise and rough ideas that serve us well enough in practical matters. The notion of a person (pudgala; puggala) or self (atman; attan) is an undeniably useful notion, and it is even one that is based upon or caused by real things in the

88

Nominalism in Pre-Diimagan Buddhism

world. It is rather like an instance of the notion of abstract understanding that we encountered above in the discussion of Ockham's nominalism (p. 78). There we saw that Ockham argued that sensation leaves a trace on the intellect, which then forms subjective abstractions that reside entirely in the mind despite being based upon external realities. In a very similar way, Nagasena readily admits that it is because of the head-hair, because of the body hair, because of the brain, because of the physical body and feelings and perception and character and awareness that the name, appellation, idea, designation, and mere label "Nagasena" can be used. If none of the above things were present, then it would be inappropriate to use the name "Nagasena," so usage of the label is not entirely groundless or arbitrary, for it is governed by constraints. But the fact remains that to the name and the idea there corresponds no one steady and reliable thing; rather, insofar as the name refers to anything at all, it seems to refer now to one thing, such as the body or part of the body, and now to another thing, such as the mind or the aspects of one's personality. Since its reference is constantly and rather unpredictably shifting, person or self is a vague and imprecise notion and cannot be regarded as corresponding to any ultimate (paramiirtha; paramattha). It is, in the language of abhidharma, a derivative idea (upiidaya prajiiapti; paiiiiatti), a technical term that we shall now investigate more fully as we turn once again to an examination of Nagfujuna's Millamadhyamakakiirikii.

3.2 Nominalism in Nagarjuna When Nagarjuna was [rrst discussed above in Chapter two, attention was drawn to his non-assertive skepticism. There it was said that Nagfujuna began his Malamadhyamakakiirikii with a famous pair of verses that can be translated as follows: I pay homage to the finest of speakers (vande vadatarh varam), who being fully awakened (sambuddhas) showed happiness (sivam deSayiim lisa) as not coming to an end (anirodham), not coming into being (anutpt'idam), not being cut off (anucchedam), not being everlasting (aSasvatam), not a single thing (anekiirtham), not many things (ananlirtham), not approaching (anagamam), not receding (anirgamam), but as dependent origination (pratltyasamutpadam), the quelling of vain thinking (prapaficopasamam).

A verse such as this is capable of serving as the basis of a. great deal of mystification. Edward Conze, for example, sees in Nagfujuna's writing in general a preoccupation with "one problem only--the conditions which

Chapter Three

89

govern the transcendental intuition of the Absolute." The members of the Madhyamaka school, says Conze, "devoted an enormous amount of ingenuity to distinguishing absolute from mere empirical knowledge, which was ipso Jacto held to be false. ,,35 Although a "transcendental intuition of the Absolute" sounds as if it might be a very impbrtannhing to have, it is not at all clear to most of us what the expression could possibly mean. It sounds rather like the fulfillment of an offer of the sort that James Joyce once reportedly made to a friend who was having trouble understanding the novel Ulysses; Joyce said "If I can throw any obscurity on the subject let me knoW."36 Another author of this century who was fond of waxing incomprehensible was D.T. Suzuki, who offers his advice on how to come to terms with the concept of Buddhist emptiness (sunyata): In order to get into the world of Emptiness, existence itself must be made to tum a somersault. One must experience sitting at the centre of existence and viewing things from this hub. Let one remain at this side of dualism and the gap between relativity and Emptiness can never be bridged. 37

Nagarjuna's favoured methods of presenting argument is characterized by both Suzuld and Conze as an example of dialectics, the form of logic that in Conze's words "maintains that all truth must be expressed in the form of self-contradictory statements. "38 Elsewhere Conze, towards the end of his long career studying Buddhist thought, wrote "There can thus be no doubt that much of Mahayana thinking can be described as 'dialectical' and that it should be interpreted as such. A study of the Tetralemma (catu$ko(i) might in addition suggest that all Buddhists depart from the Aristotelian logic in essential points .... Everywhere one moves into the space of the contradictory and the absurd, and the violation of logical laws is viewed as a small price to pay for spiritual freedom. "39 Suzuki expresses a similar sentiment when he says that once we see where the Buddhist perfection of wisdom (prajfitiparamita) stands, "we can realize why it abounds with negative phrases and irrational assertions. Its intuitions could not be expressed in any other way if they were to be expressed at all. "40 This, he explains, is why Zen masters were given to speaking and acting in ways that often strike us as odd, and would surely have struck even Nagfujuna as peculiar. Rather than explaining emptiness directly in words, the Chinese and Japanese meditation masters approached it obliquely, knowing that "even a particle of dust is not outside the mind, and when this is understood, Sun(ata and its cognate ideas will all become comprehensible. "4 But let us hope, with all due apologies to our Zen friends, that there is no harm in trying to make emptiness comprehensible through the humdrum method of speaking plainly. Rather than trying to

90

Nominalism in Pre-Diimagan Buddhism

understand a difficult passage from Nagarjuna in terms that are even more baffling than the original passage itself, at least for those of us who do not know how to make existence turn a somersault, let us begin from a point of departure different from that of· the proponents of the view that in Nagarjuna we have a practitioner of the dark science of dialectics. Let us begin by simply examining some of the key expressions in Nagarjuna's opening verses from a traditional Buddhist point of view, for each of the expressions that Nagfujuna uses in this verse has a long history in Buddhist usage that dates back to the literature of the Pali Canon. Of particular importance in Nagfujuna's opening stanzas is the pair of expressions "not being cut off, not being everlasting." These two terms have a long history that goes back to discussions between the Buddha Gautama (Gotama) and other philosophers of his day. A topic of great importance to the Indian philosophers contemporary to the Buddha was that of whether there is something, whether we call it the person or the vital principle, that continues to exist after the death of the physical organism. Some religious thinkers argued that there is something in a living being (sattva) that remains constant from the moment the living being enters the womb at conception until the death of the physical organism, and this something furthermore continues to exist after the death of the physical body. Some argued that this vital principle sought out one physical body after another. Others denied that this vital principle existed beyond the life of the physical organism. According to them the vital principle can be regarded as an epiphenomenon, that is, as an accidental by-product of the coming together of various physical elements in the form of a living being. On this theory this vital principle comes into being when sperm and egg unite in the womb and ceases to exist when the physical body that eventually evolves out of gamogenesis finally loses its breath. Proponents of this theory could also point out that it is better suited to serving as a justification for moral behaviour. If a person is eternal and unchanging, they argued, then one cannot really be killed or otherwise made to suffer, in which case it would make no real difference how we act towards other living beings. No matter how we might behave towards other beings, the true person in those beings would not be affected. On the other hand, proponents of the theory that we do somehow survive the death of the physical body claimed that the view that the person dies with the death of the body was in fact equally incapable of supporting moral behaviour. For if we are not answerable after death for the actions that we pelform while alive, then the closer we get to death, the more sense it makes to act quite selfishly and in utter disregard for those around us; but we get closer to

Chapter Three

91

death at each moment, so it follows on this view that at each moment we should become less concerned with the repercussions of what we do in the world. Solving this puzzle was therefore a matter of great consequence to religious leaders, and seekers of such leaders were constantly pressing the Buddha for his solution to the problem. His solution was quite radical. The principal weakness in both of these theories is that the vital principle discussed in each theory is supposed to remain constant throughout one's life. But whatever remains constant and undergoes no change cannot really be said to participate at all in the world of acting, changing, acquiring knowledge and so forth; it cannot act or experience the consequences of action. But such an actionless and impassible being is not at all what we in the world consider ourselves to be, for we think of ourselves as acting and being acted upon throughout the course of our existence. So the radical solution to this problem that the Buddha adopted was to reject the presupposition on which both theories rested, namely, the presupposition that there exists a self that remains stable from conception to death. This enables him to reject both of two apparently contradictory positions. The Buddha's position is summarized in this PaIi verse: If the person is destroyed when the groups [of various kinds of characteristics] are destroyed, the annihilationist view arises, which the Buddha avoided. If the person is not destroyed when the groups are destroyed, the person is everlasting like nirval]a.42

On this view, it is not the case that the person continues to exist after death, and it is not the case that the person ceases to exist after death. Nagarjuna devotes the final chapter of his Mi1lamadhyamakakarika to spelling out the full implications of the Buddha's avoiding both the view of a person's being annihilated or cut off and the view of a person's being everlasting. This is no flight from the laws of ordinary logic into dialectics or into a realm accessible only to the mystic; rather it is a simple claim that there is no person in the first place, and therefore the question of what happens to the person is one that simply does not arise. No person comes to an end, no person comes into being, no person is cut off, no person is everlasting; the notion of a person is an empty notion, because it is one that arises when we have apprehended other things that are not the person at all, things such as the physical body, feelings, perceptions, character and awareness. The concept of self, in other words, is a derivative idea. As Nagarjuna points ouJ in Mi1lamadhyamakakarika 24:18 "To be empty is to be a derivative idea (upadaya prajiiapti.)."

92

Nominalism in Pre-Diimagan Buddhism

In his exposition of the doctrine of the self or the person as a derivative idea, NagaIjuna's position is no more than a simple exposition, in almost exactly the same tenninology, of how that doctrine is handled in the Sutta Pitaka, in the Vinaya Pitaka and in the book called Kathiivatthu in the Pali Abhidhamma Pitaka. But if Nagarjuna had done nothing but expound doctrines in almost exactly the way the doctrines had already been expounded all along, it would be difficult to account for his occupying such an important position in the history of Buddhist thought. Nagarjuna did indeed make an advance in Buddhist theory by explicitly taking a number of basic suppositions in Buddhist teaching to their natural conclusions. This is especially the case in the area of Buddhist nominalism. A nominalist theory was at best inchoate in the Nikaya literature. At most we can find it said that some notions have nothing corresponding to them in the external world, and some notions are derivative ideas and therefore empty. What NagaIjuna did in his Miilamadhyamakakiirikii was to show that all concepts are derivative ideas, for all concepts are meaningful only in the context of other concepts. He showed in one family of concepts after another that no concepts if carefully examined fail to yield contradictions, and therefore the real world cannot really be at all the way we conceive it to be. Whereas we think about and understand the world in tenns of actions, actors, things affected by action, motives for acting, causes, effects, sense faculties, sensible objects, feelings, perceptions, character, awareness, elements, virtues and vices, identity and difference, self and other, the world of distress, the cessation of distress in nirviil)a, misconception and truth, NagaIjuna shows that none of these concepts is capable of withstanding the rigours of principled thinking (yoniso manaskiira). All are derivative ideas, and there are no primitive ideas upon which all these derivative ideas are based. In the final analysis the ordinary world defies our understanding. Whereas we saw it said in the Potthapiida Sutta that the Buddha could use folk expressions such as "self' or "person" without being led by using such expressions into desire, conceit and opinions, Nagarjuna extends that immunity to the use of all language. When the unreflective person uses language, he is supposedly seduced into believing that he is actually saying something about something. The wise person is not so beguiled. 43 A fully developed nominalism becomes for Nagarjuna a means of protecting oneself against the opinions (drs(i; diUhi) that are implicit in language and

Chapter Three

93

that if not seen in their true light can stand as obstacles to inner peace or nirvar;a.

3.3 Nominalism in Vasubandhu One final Indian Buddhist philosopher whose contributions to the development of a Buddhist theory of nominalism must be taken into account is Vasubandhu, whose work is directly relevant to the study of Dinnaga's philosophy in that Dinnaga devoted himself explicitly to the task of expounding, and sometimes criticizing, Vasubandhu's ideas. Vasubandhu was one of the most prolific and influential authors in scholastic Buddhism, and his works covered a wide range of topics from abhidharma to the theory of debate and the art of persuasion. Three specific areas in which Dinnaga's thought is built upon a foundation laid down by Vasubandhu are abhidharma, theory of debate and the development of a scholastic trend that may be called Buddhist phenomenalism. The theory of debate is not one that has much relevance to the subject matter of this book, so in the following discussion of Vasubandhu I shall focus on two themes that emerge in his great treatise on abhidharma and in his phenomenalistic writings.

3.3.1 Vasubandhu's theory of two truths In the opening paragraph of his excellent survey of various manifestations of the theory of multiple levels of truth in Buddhism, La Vallee Poussin correctly points out that it is a universal preoccupation of philosophers everywhere in the world to try to sort out those statements that pass rigorous conditions of truth from those statements that simply pass as true in everyday life. 44 It was certainly a matter of great interest to Buddhists from the very dawn of the Buddhist movement, and indeed one could say without exaggeration that it is the central issue in Buddhism as a way of life to distinguish somehow between all the things we believe in virtue of uncritical acceptance of what is commonly said by ordinary people on a daily basis and the things we believe, or should believe, in virtue of principled thinking (yoniso manaskara). In practically everything from the arena of Buddhist philosophy that we have examined up to this point we have been seeing instances of this central issue. What I hope to do now is to sketch out some of the explicit discussion on how this

94

Nominalism in Pre-Diimagan Buddhism

distinction between what is usually called conventional truth and ultimate truth is to be made. La Vallee Poussin has discussed the evolution of the Pali term "sarhmuti" or "sarhmati" into the incorrectly Sanskritized "sarhvrti" and all

the philosophical implications of this false Sanskritization. But since his treatment of the subject may not be widely available, let me just recapitulate what he reported on that matter. In the Pali tradition the term "sarhmati," which is derived from the verb root MAN (to think), suggests that which is commonly thought or believed; "sarhmutisacca" is a technical term that signifies the "conventional truth," or that body of beliefs that the general population takes to be true. The conventional truth is the set of beliefs that are implicit in our daily way of talking about things as well as in our daily behaviour. Stock examples that are given are such things as the fact that people ordinarily bow down before kings and other powerful human authorities and use elevated terms of address such as "Lord" in speaking to them and about them. In this respect kings and powerful people are treated in the same manner as the popular gods and can therefore be regarded as gods by popular convention (sarhmati-deva). Similarly inanimate objects are often personified in daily speech, such as when we speak of a tree living and dying despite the fact that the principle of life in plants is different from that in animals and therefore plants and animals cannot be said to live and die in quite the same senses. 45 Similarly, but of much greater importance in the quest for nirvalJa, there is in ordinary speech an implicit belief in a self or a person. But this is a vague idea and one that will not stand up under close examination, so all talk about a self or soul or person is also classed by the Pali Buddhists as conventional belief but not as "rigorous truth (paramiirtha; paramatthasat).,,46 The rigorous truth is that set of opinions that can stand up under close scrutiny and therefore can be regarded as accurately representing things as they really are as opposed to things as we commonly assume them to be. Vasubandhu's work on abhidharma became the standard work on the subject for most of the Sanskrit-speaking Buddhists of the northern part of the Indian subcontinent and of central Asia and of those parts of the eastern Asia that were directly influenced by the intellectual achievements of the academic Buddhism of the northern subcontinent. In the seventh chapter of his Abhidharmakosa Vasubandhu discusses the various types of cognition (jfi.iina). Although he acknowledges that there are many classifications into which cognitions can be placed, there are basically only two types. On the one hand there are cognitions that are accompanied by contaminants (iisrava), and on the other hand there are cognitions that are

Chapter Three

95

uncontaminated. 47 The notion of contamination has the general sense of that which keeps one bound to and confused by the world of rebirth and distress. Traditionally four types of contaminant are enumerated: contamination through sensual pleasure (kama), contamination through birth and living (bhava), contamination through misconception (avidya), and contamination through opinions (drsti). The conventional truth is that which is contaminated by these things, while the rigorous truth is free of such contamination. 48 The subject matter (visaya) of a conventional cognition can be anything whatsoever that is commonly believed by the ordinary populace to exist. Examples of such things that Vasubandhu offers are water jugs, cloth, females, males and so forth. Cognitions that focus upon such subject matter as macroscopic objects, such as men and women or persons and selves or jugs, cloth, pieces of furniture and chariots, are not only contaminated by attachments to sensual enjoyment and so forth but are dealing with objects that can be shown in the final analysis not really to exist at all. Insofar as such objects exist at all they can be said to exist only by popular consent. But if one were to make a catalogue of all the things that really do exist in the world, as opposed to a catalogue of things that occur as the subject matter of human thoughts about the world, there would be no water jugs, pieces of cloth, tables, chairs or chariots. But if none of these things truly exist, what is it that is left over? This is one of the questions that Vasubandhu takes up in his chapter on people who follow the path of Buddhism. In his discussion of what it is that distinguishes the conventional truth from rigorous truth, Vasubandhu offers a verse that states "That is conventionally real of which there is no perception when it is broken like a water-jug and when like water it is abstracted by the intellect from other things. The rigorously real is otherwise. ,,49 This verse happens to contain the expression that came to be so important in the nominalism of Diimaga and DharmakIrti, namely· the expression for the concept of abstraction from others (anyapoha), or mentally separating an object from the context in which it is experienced and associating it with similarly abstracted objects in the recollection of previous experiences. In his own prose commentary to his verse; Vasubandhu explains the passage as follows. That is conventionally. real of which there is no perception when it is broken into parts. An example is a water-jug, because when that is broken into shards there is no perception of a water-jug. And that should also be understood as conventionally real of which there is no perception when one has mentally sorted other properties out. An example is water, because when one has mentally sorted such properties as material form out, there is no perception of water. But conventional designations are applied to those very things, so one who says on the authority of convention that there is a water-jug and there is water is

96

Nominalism in Pre-Dinnagan Buddhism speaking the truth rather than a falsehood. And. so this is a conventional truth. The rigorously real is different from that. That is rigorously real of which there does arise a perception even when it is broken and even when there is mental abstraction from other properties: An example is material form (Tiipa), because when that object is broken into atoms and even after sensible properties50 are sorted out by the intellect, the perception of the essence (svabhtiva) of material form does arise. Feelings can be viewed in the same way.51

Apparently Vasubandhu's intention here is to define the ultimately real or the rigorously real as an absolute simple, that is, a thing that cannot be physically divided into smaller parts or a thing the concept of which cannot be reduced to more primitive concepts without losing its essential nature. As we saw above (p. 73), a similar notion of what it is to be fully real was found in the work of Plato, where the fully real was seen as the absolutely singular that could not be broken down into smdler physical parts. A view such as this one leads to a number of difficulties, as Aristotle showed (see p. 75 above), for if it is taken to its logical conclusions, then either only one thing is fully real, namely Being itself, or only absolute individuals such as atoms are real. The tendency of early Buddhist abhidharma was towards the view that only absolute individuals are rigorously real, a view that amounts to a kind of atomism. Any complex thing that is composed of absolute simples is real only in a conventional sense; that is, it has reality only insofar as a perceiving mind focuses attention upon it in an essentially arbitrary manner and segregates it from the total field of experience. Any word or phrase, therefore, that applies to a complex being of this kind names a concept and not a real item in the world. But since names are given only to the macroscopic complex beings with which we deal in daily life--we do not, in other words, give a unique name to each and every atom in the universe--it would seem to be most consistent with the ontology of Buddhist abhidharma that all words apply, as in Ockham, only to "intentions of the mind."

3.3.2 Vasubandhu's phenomenalism According to Buddhist tradition Vasubandhu composed the verses of the Abhidharmakosa from the point of view of the V aibha~ika school, then composed a prose commentary from the point of view of the Sautrantika school, and later in his life composed a number of treatises from the point of view of the Yogacara school. According to later Indian tradition, the Vaibha~ika and Sautrantika schools belonged to' the Humble Path (HInayana) of Buddhism, while the Yogacara school is representative

Chapter Three

97

of the Grand Path (Mahayana). Some Buddhists and some historians have placed much importance on the distinction between these two paths or vehicles and have therefore found it significant that Vasubandhu underwent a "conversion" from the one path to the other, as if these changes of schools must have been the outcome of dramatic changes of outlook. Part of the reason behind the assumption that a change from the Humble to the Grand Path must have represented a somewhat radical change in outlook is undoubtedly that the literature of the Mahayana is so often of a polemical nature and spares no efforts in showing the inadequacies of the arhants who figure so prominently in the Buddhist Agama literature. Indeed it is difficult to read some Mahayana siitra literature without thinking that the followers of the new Grand Vehicle had nothing but contempt for the allegedly petty-minded and hopelessly stupid heroes of the earlier Buddhist literature, and it is rather easy to imagine that taking the Mahayana seriously might have entailed a supreme dissatisfaction with the traditional teachings of the Buddha Gautama and his earliest followers. And a further reason behind the assumption that it was a big step from one path to the other is that later historians of Indian Buddhist thought, in their efforts to show the salient differences among the various philosophical trends, often exaggerated those differences and failed to draw attention to the important similarities among the diverse schools of Buddhism. Perhaps they took those similarities for granted, and perhaps they wrote at a time when Buddhism as a living reality was all but forgotten in India, but whatever the reasons, there is no doubt that the differences among the schools did come to be far more celebrated than the similarities. Many modern scholars, especially those who wrote at a time when our knowledge of classical Indian thought was rather sketchy and heavily influenced by late medieval and modern presuppositions about the character of philosophies of former ages, followed the traditional line of seeing rather stark differences between the Sautrantika position and the Yogacara position. Examples of this trend of scholarship are Stcherbatsky, S. Mookerjee, D.N. Shastri, and A. Singh. Singh, in fact, finds the differences between the Sautrantika and the Yogacara so great that he finds it difficult to believe that the author of the Sautrantika Abhidharmakosa could have been the same person as the author of the Yogacara works, and he argues that there must have been two authors who happened to have the same name. 52 The key philosophical issue that is supposed to separate the Sautrantika and the Yogacara perspectives is that of whether or not the objects of experience actually exist independently of our awareness of

98

Nominalism in Pre-Dirinagan Buddhism

them. One representative account of how these two schools are supposed to differ on this issue should suffice to illustrate the point under dispute. The account given by D.N. Shastri is clear and worth quoting in extenso. The Sautrantika system of thought is an example of what Shastri calls representationism, which he describes as follows: According to this theory, external objects are not apprehended directly and immediately, but through the cognitions of these objects. The objects transfer their forms to their cognitions, and the cognitions, having thus acquired the forms of the external objects, become their representatives. We have thus a representative perception of objects, and not a direct one. Hence the theory is called representationism. External objects, not being perceived directly, are only inferred from their cognitions to which they impart their forms. Orthodox Indian writers, in their compendia of philosophical systems, have ascribed this theory to the Buddhist Sautrantika schooL53

If this Orthodox Brahmanical account of the Sautrantika school is essentially correct, then that school holds a view rather similar to the theory of John Locke in European philosophy, as Shastri himself points out. In contrast to this representationist view is the view that is traditionally called, in European philosophy, subjective idealism. Shastri describes it as follows: Subjective idealism consists in the assertion that there are no other things than thinking beings; that the things we believe ourselves to perceive are only the ideas of thinking beings. In short, the theory holds that there is no objective world independent of the perceiving mind .... In Indian philosophy it is represented by the Yogacara school of the Buddhists. In western thought, Berkeley is the chief representative.54

Shastri points out in a note that the ascription of the Sautrantika school to a representationist view is not found in any Buddhist works but occurs in several Brahmanical treatises, and later on he sounds the cautionary note that it is "obvious that the orthodox writers of the later part of the postBuddhist period are superficial, confused and even mistaken in their account of the Buddhist systems. "55 Many specialists in Indian Buddhist philosophy have apparently come to have the same wariness towards Brahmanical accounts of Buddhist philosophical schools, for there has come to be a general movement away from portraying the Yogacara school as an instance of an idealistic form of philosophy. Examples of this nonidealistic interpretation of Yogacara are Alex Wayman (1979), Thomas Kochumuttom (1982), and Bruce C. Hall (1986). In the description of Vasubandhu's version of Yogacara thought, I shall follow a line of interpretation somewhat more like that of these more recen~ scholars than that of earlier scholars who saw in Yogacara a kind of subjective idealism.

Chapter Three

99

Vasubandhu begins his treatise called Virhsikiivrtti with this statement: In the Mahayana it is taught that what is derived from the three elements is nothing but phenomena, because the Sutra says "Sons of the Victor; whatever is derived from the three elements is nothing but thought." Thought, mind, awareness and phenomenon are interchangeable terms. What is meant in this context is thought along with what is connected with it. The phrase "nothing but" is for the sake of ruling out objects.56

This passage is full of technical terms and requires a certain amount of explanation before it becomes clear what is being said. The first term that requires some explanati~~' is the term "phenomenon (vijflapti)." Morphologically the term is derived from the verb "vijiiniiti," which means one is aware. The causative form of the verb has the sense of maldng someone else aware or informing someone, and the abstract noun "vijflapti" formed from the causative verb root has the sense of information. It is, therefore, that to which one's attention is drawn or that of which one is aware. There are a number of philosophical terms in Sanskrit that can be used in the sense of that of which one is aware. Many of them carry at least a suggestion that the thing about which one is aware is somehow distinct from the action of awareness itself and is more or less passively received by or reflected in consciousness. But Vasubandhu is intentionally questioning the assumption that the correct account of experience is that a passively conscious subject experiences directly something entirely outside the conscious subject itself and is suggesting instead that what there really is is simply an integrated experience onto which we project (or out of which we abstract) the notions of perceiving subject and object perceived. Since Vasubandhu wishes to talk about the ordinary in an extraordinary way, his task is simplified greatly by spealdng in a language that is not laden with the very implicit assumptions that he is trying to question. Therefore, he employs a term that is philosophically relatively neutral in that it does not imply what is external to awareness but can be pressed into service to refer to information from the point of view of its already being within awareness. The Greek term "phainomenon" is a present participle formed from the verb "phaino" literally meaning to shine and by extension to be in evidence or to appear. A phenomenon is an appearance in perception. The term is used to name the subject matter of an experience as it is perceived, in contradistinction to the objective cause or the stimulus that is responsible for bringing the perception about. Since the term vijflapii seems to be used by Vasubandhu in a similar way, it can be translated tolerably accurately by the term "phenomenon." This is the

100

Nominalism in Pre-Dinnagan Buddhism

translation suggested by B.C. Hall, who offers this neatly articulated summary of the philosophical advantages of adopting such a translation: The tenu vijfiapti signifies a "phenomenon" of consciousness, a "manifestation" to consciousness, or a "percept" --so long as one bears in mind that these tenus should not be taken in a naively realistic or a naively idealistic sense....To translate vijfiapti here by "representation" conveys its "public" aspect, but seems to imply representation of something, presumably an external object or referent, which suggests a "representational" theory of knowledge. On the contrary, the purpose of the argument through'out the Vilhsatikii is to show that the concept of vijfiapti suffices to make sense of perception and that the concept of an external referent (artha) is logically superfluous.57

The motivation behind declaring that all experience is nothing but phenomena is, according to Hall, not to make a "metaphysical assertion of a transcendental reality consisting of 'mind-only.' It is a practical injunction to suspend judgment: 'Stop at the bare percept; no need to posit any entity behind it."158 A careful reading of Vasubandhu's Yogacara writings, I think, supports Hall's interpretation quite well. The next term in need of some explanation is "element (dhiitu), the most basic sense of which is a primary element such as earth, water, fire or wind. By extension it means the sensible properties by which the elements are experienced, and by further extension it means any elementary component into which the totality of our experience can be analysed. The specialists in meditation as well as in abhidharma factored experience into eighteen elementary components: the six kinds of sense datum, namely, colour (rupa), sound (sabda), smell (gandha), taste (rasa), touch (sparsa) and abstract property (dharma); the six sense faculties that are specialized in sensing those data, namely, the eye, ear, nose, tongue, body and mind; and the six resulting types of act of awareness, one corresponding to each of those kinds of sense datum. The early Buddhists also discussed two "worlds" of experience, by which was meant two modes of experiencing the world. The mode of the foolish masses of ordinary people was called the world of matter (rupa-Ioka), since the dominant feature of the .experience of the unreflective masses of people is the desire to enjoy material things that are enjoyable and the desire to avoid what is tedious or indifferent. The mode of the thinker and contemplative was called the nonmaterial world (arupa-loka). The contemplative or philosophical nonmaterialistic way of experiencing the world was viewed as necessary but not sufficient for getting rid of all cravings and attachments and attaining nirviif}.a. Presumably the reason for the non-material mode's not being regarded sufficient for complete non-attachment was that it is possible to be non-attached to material things but still quite attached to such non-

Chapter Three

101

material things as ideas, ideologies, knowledge, dogmas and other abstract things. Thus the entry into the non-materialistic way of experiencing was a kind of intermediate stage between the world of materialism and the world of non-attachment or nirvalJ.a. This view of the three worlds of experience is found in such early Buddhist texts as Sutta Nipata 754-755: . Those beings who are attached to material things, and those who abide in non-material things without experiencing a cessation [of attachments], are bound for rebirth. But those people avoid death who, being well-established in the non-material after fully understanding the material, are released in the cessation [of attachments].59

In other passages in Buddhist Nikaya literature a distinction is made between the material world as experienced by the meditator and the material world as experienced by the unreflective person of desire. Thus the material world (rapa-taka) was divided into the world of desire (kamalaka) and the world of a pure experience of matter, in the sense of an experience uncontaminated by desire. This latter pure experience of the desireless meditator came eventually to be the only world to be commonly called by the name rapa-taka. The non-material world then came to be associated with four specific meditational practices, namely the contemplations of unlimited space, of unlimited awareness, of nothing at all and of being neither aware nor unaware. 60 With regard to each of these three-world schemes, one can still talk of the elementary components (dhtitu) that go into making up the experience. One can speak of the elements of the materialist world-experience in the telescoped phrase "materialistic elements (rapa-dhtitu)" and so on with the other worldexperiences. The totality of elementary components in all three types of world-experience can be referred to by the abbreviated phrase "the three elements (tridhatu)," and the experienced world that is derived from them is called "what is derived from the three elements (traidhatukam).,,61 In effect what is meant is the totality of all experience by all beings.

Whichever of these three-world schemes one is following, the underlying supposition is that the world we experience is principally subjective. A notion that virtually every Buddhist would accept is that the experiences each of us has of the world are more a function of our own mental predispositions than of any objective realities. What makes one person's experience of a given set of objective circumstances different from another person's experience of those circumstances is obviously not the circumstances themselves, but rather the moods, emotional presentiments, memory associations and so on that each person uniquely brings to the circumstances. To say only this much is hardly to be a subjective idealist. The question to which we must turn, then, is whether

102

Nominalism in Pre-Diimagan Buddhism

Vasubandhu was making a claim stronger than the rather uncontroversial claim that private and subjective factors playa very great role in the kinds of experiences we have. Indeed, Vasubandhu appears to be making a stronger claim than that when he says "what is derived from the three elements is nothing but phenomena" and "the phrase 'nothing but' is for the sake of ruling out objects (artha)," by which it seems most likely that he means that the objective component of experience is being excluded from consideration, although it is hard to say for sure without further investigation, since there is probably no term in Sanskrit more rich in meaning and therefore more vague than the word "artha." Let us go on, then, to see how Vasubandhu argues his case. The reason that Vasubandhu adduces for claiming that all experience is nothing but phenomena is that phenomena appear in the absence of an objective basis. For example it is possible for people to have the experience of vision even in the absence of an external colour acting as the objective counterpart to the perceived colour that is the subject matter of the experience. People who have defective sense faculties may have perceptions that do not correspond to reality, and people who are asleep have dreams. Such experiences are commonplace. In addition, one can cite the fact that there are beings who have the experience of being tormented in hell, where they are subjected to all manner of unpleasant sights and sounds and bodily sensations. It stands to reason that these experiences of the torments of hell are purely subjective and without objective counterparts, because the reports of hellish experiences speak of guardians and beasts that prowl environments of heat so intense as to be unbearable. These guardians and beasts reportedly torment the inhabitants of these hellish regions. But reason shows that these guardians and beasts of the hells could not themselves suffer from the conditions that the victims of hell find intolerable, for if they suffered as much as the victims they would be so overcome with pain as to be incapable of performing their tasks of tormenting the victims. Therefore it ml,lst be the case that these guardians and beasts, and indeed all the other torments of hell, are purely subjective experiences that no matter how intense cannot have an objective counterpart in the external world. So, taking all these instances into consideration in which there is a vivid and intense experience without there being an artha or objective counterpart, it can be seen that not all experience requires an external stimulus. So far this is an innocuous claim, one that hardly anyone would dispute. Once it has been established that we all have experiences that do not correspond to external realities, the natural question that arises is: When our senses are capable of malfunctioning by presenting us with experiences of things that are not really. there, is there anything that we can rely upon to help us correct the defects of the senses?

Chapter TIrree

103

The answer that most readily comes to mind is that there is indeed something we can rely upon, namely, the intellect. We know that some of our experiences ani not accurate, because they do not stand up to reason; the argument concerning the hellish experiences gives one example of t.he way that we rely on reason to override the presentations of the senSes. But if reason is the final arbiter in deciding whether the presentations of the senses can be trusted, reason is also the final arbiter in deciding whether the presentations of reason itself can be trusted. Reason, for example, must be the source of the notion of atoms (paramiilJu), since atoms are too small to be directly sensed through vision or any of the other senses. Atomicity is a concept through which we make an attempt to understand the world, and it has every appearance of being a most natural sort of concept, for prima facie there is nothing at all odd in the notion of that which is tiny (alJu) to the greatest possible extent (parama). But a little reflection on the matter shows that the concept of atoms leads to inconsistencies. Vasubandhu deals with this topic in verses 11-14 of the VirhSikii. The argument that he gives there can be condensed as follows. If an atom is really ultimately tiny, then it cannot be made up of parts that are physically smaller than the atom itself. But if the atom has no parts, then the region of space occupied by the atom can have no subregions; there can be no upper region or lower region or eastern region and so on. But if that is true, then the atom can occupy no space at all. And if one atom occupies no space, then several atoms added together cannot occupy any space. Therefore what reason presents to us, namely, that a macroscopic object is a complex of a very large number of ultimately small component parts, cannot really be the case after all. The atom, then, is a concept to which there cannot correspond anything in the real world. If we now look back at the conclusions about the nature of the rigorously real that Vasubandhu set down in the Abhidharmakosa and compare them with the results of his reasoning in the Vijfiaptimatratiisiddhi, we may be able to gain some appreciation of how his ideas taleen all together might have informed the thinking of Diimaga, who after all inherited all the works ofVasubandhu. As we saw in the discussion of the rigorously real in Abhidharmakosa, no complex thing that can be reduced to more primitive parts is rigorously real. Any name given to such a complex object therefore names only a concept and not an item in the real world. It was also suggested that the tendencies of the ontology expressed in the Abhidharmakosa were towards a kind of atomism. But in the Vijfiaptimatratiisiddhi we saw that a correction is made to avoid the absurdities of an atomistic view of the world. In this latter text it is argued that even absolute simples such as atoms are no more than conceptual

104

Nominalism in Pre-Difmagan Buddhism

fictions. And from this it would follow that any name given to-the concept of an absolute simple--a word such as "atom (paramanu)"--also names only a concept and not an item in the real world. But if every word that names a complex thing in fact names only a concept, and every word that names an ultimately simple also names only a concept, it follows that any word that names anything at all must name only a concept, for there can be nothing besides what is simple and what is complex. Whatever else Vasubandhu's philosophy can be called, it'is surely a form of thoroughgoing nominalism, a kind of nominalism like that found in the LaIikavatara Sutra, which says: Personal identity, continuum, groups, causal conditions, atoms, primordial matter, and God the creator are regarded as mere ideas. 62 And like the nominalism of Nagfujuna and earlier Buddhists, Vasubandhu's nominalism serves as a safeguard against the developments of the opinions implicit in language that if not guarded against stand as obstacles to nirvana. That this is the motivation behind his nominalism is clear in his concluding verses to the Trirhsika Vijfiaptikarikii portion of his Vijfiaptimiitratiisiddhi: When awareness apprehends no objective basis at all, then it is stationed in the fact that there is nothing but awareness, because in the absence of an apprehensible object there is no apprehension of it. That nonapprehension is without thought and that cognition is superior to those of the ordinary populace. 63 Turning away the objective support is of two types, because it rejects what is wrong. It alone is an element without contamination, unthinkable, healthy, stable. This is the jo~ous mass of liberation. 64 It is called the Dharma of the great silent sage. 5

-- Notes --

Chapter 3. Nominalism in Pre-Dinnagan Buddhism 1. Fowler 1914 tr., p. 19. 2. Fowler 1914 tr., p. 21. 3. Cooper (1941 tr., p. 174) renders "eidos" by "ideal form" in his translation of the Euthyphro.

Chapter Three

105

4. Republic 596. Shorey 1930 tr, p. 820. 5. lowe this summary 'of Plato's theory of knowledge to Copleston 1946a, pp. 173-178, whose lucid discussion I am abridging and simplifying considerably. 6. Metaphysics 980a in Ross 1941 tr., p.689. 7. A discussion of the technical terminology used by Aristotle is found in Richard Hope's translation of the Metaphysics. Encoded into this translation are frequent indications of the exact terminology used. I have profited in this discussion of Aristotle's criticisms of Plato by comparing three translations: Tredennick 1933 and 1935, which also contain the Greek text, Ross 1941, and Hope 1952. All indications of the standard Latin translations of Aristotle's Greek that occur in my discussion were taken from the glossary of Hope's translation. 8. A similar discussion occurs in Posterior Analytics 2: 19, where Aristotle states his conclusion thllt our knowledge of universals is not something innate in the soul that is recovered or recollected through experience, as Plato had argued; rather, general knowledge is acquired after seeing particulars through sense-perception, since the human soul has a natural capacity to compare present sense impressions with former ones and to systematize them. 9. Metaphysics Gamma 1003a in Ross 1941 tr., p. 731-732. 10. Metaphysics Alpha 990b in Ross 1941 tr., pp. 706 ff. An account of Aristotle's criticisms of Plato's theory also appears in Copleston 1946b, pp. 35 ff. 11. Metaphysics Alpha Minor 994a in Ross 1941 tr., p. 7l3. 12. Metaphysics Beta l001a in Ross 1941 tr., p. 727-728. l3. Metaphysics Beta l003a in Ross 1941 tr., p. 731. 14. Metaphysics Zeta 1034a in Ross 1941 tr., p. 795. 15. Gilson 1955, p. 487. 16. Loux 1974 tr., p. 79. 17. Loux 1974 tr., p. 81. 18. An account of Ockham's views on how the intellect acquires abstract notions appears in Gilson 1955, p. 495. 19. Warder 1970:164. 20. Andersen and Smith (19l3 ed., pp. 117-118)

106

Notes

TesartJ vo 'ham vyakkhissartJ Vasettha ti Bhagava anupubbartJ yathatathartJ jativibhailgartJ palJanartJ, aiiiiamaiiiia hi jatiyo. TilJarukkhe pi janatha, na capi patijanare, liilgartJ jatimayartJ tesartJ, aiiiiamaiiiia hi jatiyo. Tato kIte patailge ca yava kunthakipillike, liilgartJ jatimayartJ tesartJ, aiiiiamaiiiia hi jatiyo. 21. This argument is found in Sutta Nipata Viiseuhasutta verses 607-619. The text of this sutta, also entitled ViiseUhasutta, also occurs as sutta 98 of the Majjhima Nikaya, volume 2.

22. Andersen and Smith (1913 ed., p. 119) Paccattam sasarIresu manusesv-etartJ na vijjati vokaraii ca manussesu samiiiiaya pavuccati This rather terse statement is difficult to translate gracefully. Other attempts have expressed generally the same sentiment as mine and should be looked at for the purposes of comparison. FausbolI (1881 tr., pp. 110-111): "Difference there is in beings endowed with bodies, but amongst men this is not the case, the difference amongst men is nominal (only)." Norman (1984 tr., p. 104): "This (difference) is not found individually among men in respect of their bodies, but among men difference is spoken of as a matter of designation." Horner (1957 tr., p. 382): "Nothing unique is in men's bodies found: The difference in men is nominal." 23. The Pali and Sanskrit word 'liiti" most literally means birth, as does the Latin "natura." Eventually, as we shalI see in subsequent chapters, "}iiti" becomes a philosophical technical term that comes very close in meaning to what in European philosophy is called a "universal." 24. Oldenberg 1929 ed., p. 13: "riipartJ bhilckhave anatta, rilpaii ca h' idartJ bhikkhave atta abhavissa, na yidartJ rupartJ abadhaya sartJvatteyya, labbhetha ca rupe evartJ me rilpartJ hotu, evartJ me rupartJ rna ahosIti. yasma ca kho bhikkhave rilpartJ anatta, tasma rilpartJ abadhaya sartJvattati, na ca labbhati rupe evartJ me rupartJ hotu, evartJ me rupartJ rna ahoslti." 25. A good review of the Indian Buddhist literature on the question of the self appears in Krishan 1984, where there is also some discussion of what modern scholars have said about the philosophical difficulties involved in the traditional Buddhist position. An article that contains a good deal of information on the controversy among modern scholars on how to interpret the early Buddhist position on the self is Ching 1984. 26. Davids and Carpenter 1890 ed., p. 182: "Tassa ya purima kama-saiina sa nirujjhati. Vivekaja-pItisukha-sukhuma-sacca-salJlJa tasmirtJ samaye hoti, vivekaja-pItisukhasukhuma-sacca-saIJIJI yeva tasmirtJ sam aye hoti. Evam pi sikha eka saiina uppajjanti, sikkha eka saiina nirujjhanti." 27. Davids and Carpenter 1890 ed., p. 185: "Yatha yatha kho Pott!Japada nirodhartJ phusati, tatha tatM 'hartJ saiiiiaggartJ paiinapemi, evartJ kho ahartJ Pott!Japada ekartJ pi saiiiiaggartJ paililapemi, puthu pi saililagge paililapemIti." " 28. Davids and Carpenter 1890 ed., p. 185: "Kim pana tvam Ponhapada attanartJ paccesi?"

Chapter Three

107

29. Davids and Carpenter 1890 ed., p. 191: "Sabbe va kho ete Potthapada paribbajaka andha acakkhuka, tvam yeva nesam eko cakkhuma, ekailsika pi hi kho Potthapada maya dhamma desita paiifiatta; anekailsika pi hi kho Potthapada maya dhamma desita pafifiatta." 30. Buddhaghosa (1931 ed., p. 381) glosses "mogha" with "tuccha," which means empty or deserted. He then says that the intention of this statement is that when one pays attention to the oneself as physical body, the self as mind-stuff stops being present. 31. Po{{hapiida Sutta 53. Davids and Carpenter 1890 ed., p. 202. 32. Davids and Carpenter 1890 ed., p. 202: "ltima kho Citta loka-samafifia loka-niruttiyo loka-vohara loka-pafifiattiyo yahi Tathagato voharati aparamasan ti." 33. Buddhaghosa 1931 ed., p. 383. 34. Trenckner 1880 ed., p. 25: "Nagaseno ti kho aham maharaja fiayami...api ca matapitaro namam karonti Nagaseno ti va Siiraseno ti va VIraseno ti va SIhaseno ti va, api ca kho maharaja sankha samaiifia paiifiatti voharo namamattarh yad-idam Nagaseno ti, na h'ettha puggalo upalabbhauti." 35. Conze 1962:239. 36. Reported in David Kelly, "Go playful into that good night," The New York Times Book Review, January 18, 1987. p. 15.

37. Suzuki 1953:274. 38. Conze 1962:261. 39. Conze 1980:45. The excessive dread that Conze held of Aristotle in general and the law of contradiction in particular, which Conze argues was almost imposed upon the Greeks by Aristotle's fiat and displaced a much more spiritually awakened way of thinking in pre-Aristotelian Greek thought, is but one instance of the phenomenon, which at one time had widespread occurrence among Western students of Buddhist thought, of panic in the face of rationality. Just why this was so would make an interesting study in itself and would reveal much about the history of the rather distorted and romanticized view of Buddhism that still prevails among some intellectuals and among many antiintellectuals in the West. 40. Suzuki 1953:277 41. Suzuki 1953:292 42. Kassapa 1961 ed.,p. 39. khandesu bhijjamanesu, so ce bhijjati puggalo uccheda bhavati ditthi, ya buddhena vivajjita khandesu bhijjamanesu, no ce bhijjati puggalo puggalo sassato hoti, nibbanena samasamo ti

108

Notes

43. Suzuki (1953:261) offers this summary of the gist of Nagar-juna's philosophy: "According to Nagar-juna, all dharmas are endowed with these characters: existentiality, intelligibility, perceptibility, objectivity, efficiency, causality, dependence, mutuality, duality, multiplicity, generality, individuality, etc. But all these characterizations have no permanence, no stability; they are all relative and phenomenal. The ignorant fail to see into the true nature of things, and become attached thereby to the idea of a reality which is eternal, blissful, self-governing, and devoid of defilements. To be wise simply means to be free from these false views, for there is nothing in them to be taken hold of as not empty." 44. La Vallee Poussin (1937:159) says "L'importance du probleme des deux verites, verite de samvrti au d'apparence, verite de paramartha au verite absolue, est grande dans Ie Grand Vehicule, dans Ie Vedanta, dans la speculations indienne en general, et, on peut dire, dans la philosophie universeIIe." 45. In classical Indian Buddhist theory of life, plants were not regarded as sentient beings. But sentience was seen as a defining attribute of life in the strictest sense of the term, so that trees and other insentient forms of "life" could be called living only in a metaphorical sense. There were philosophers in India who argued, on the ground that plants turn to face the sun and put forth roots in search of water and so forth, that plants too are sentient. Most Buddhists rejected this theory, countering that this movement of plants can be explained in mechanical terms and need not be explained by appealing to the principle of sentience, whereas in animals the principle of sentience is required to account fully for their behaviour. 46. La Vallee Poussin 1937:160. 47. The term "asrava" is one of the most difficult technical terms in Buddhist thought to translate smoothly into English. Among the translations that have been used for this technical term by English translators are "depravity," "impurity," "intoxicant" and "outflow. II

48. Abhidharmakosa 7.2. In Vasubandhu 1967 ed., p. 391. 49. Abhidharmakosa 6.4. In Vasubandhu 1972 ed., p. 889.

yatra bhinne na tadbuddhir anyapohe dhiya ca tat gha\limbuvat samvrtisat paramarthasad anyatha 50. La Vallee Poussin (1923 tr. vol. 4, p. 141) seems to render the phrase "rasar-Mn dharman apohya buddhya" as if it read "rasadin dharman apohya buddhya," for he translates the phrase as "en retirer par l'esprit la saveur et les autres dharmas." Perhaps the error lay in Xuanzang's translation into Chinese, which La Vallee Poussin was following since the Sanskrit text of Abhidharmakosa had not yet been rediscovered when he made his French translation. I have taken the Sanskrit text as it stands, construing "rasarha" in the sense of fit for experience, or capable of being experienced, or sensible. The term "rasa" is quite often used in PaIi and Buddhist Sanskrit in the sense of experience in general rather than in the sense of flavour in particular, and I see no reason why in this context special attention would be drawn to taste (saveur) as opposed to any other sensible property.

Chapter Three

109

51. Vasubandhu 1972 ed., p. 890: yasminn avayavaSo bhinne na tadbuddhir bhavati tat sarhv[tisat. tadyatha ghatal:t. tatra hi kapaIaso bhinne ghatabuddhir na bhavati. yatra canyan apohya dharman buddhya tadbuddhir na bhavati, taccapi samvrtisad veditavyam. tadyatha ambu. tatra hi buddhya rupadin dharman apohyambubuddhir na bhavati. te~v eva tu samvrtisamjiia lqteti sarhV!1ivasat gha(li:mbu castTti bruvantal;l satyam evahur na m'$a. ity etat sarhvrtisatyam. ate anyatha paramarthasatyam. tatra bhinne'pi tadbuddhir bhavaty eva. anyadharmapohe 'pi buddhya tat paramarthasat. tadyatha riipam. tatra hi paramaIJubhinne vastuni rasarhan api ca dharman apohya buddhya rupasya svabhave buddhir bhavaty eva. evam vedanadayo 'pi dra~tavyal;l. 52. This view is argued in Singh 1984. An earlier version of a theory that there were two Vasubandhus was put forward in several works by Erich Frauwallner (1951, 1958, and 1961), but the basis of Frauwallner's hypothesis was that the Buddhist tradition records two separate dates for the life of Vasubandhu. In order to reconcile this controversy over the date of Vasubandhu's life, Frauwallner suggested that there were in fact two great Buddhist masters by the same name who lived at different times. The later of these two Vasubandhus, according to Frauwa11ner, composed both the Abhidharmakosa and some of the key works of the Yogacara school. For an account of the key differences between Frauwa11ner's two-Vasubandhu hypothesis and Singh's, see Hayes 1986c. 53. Shastri 1964:41. 54. Shastri 1964:42-43. 55. Shastri 1964:59. 56. An edition of the Twenty Stanzas (Virizsikiivrtti) of the Vijiiaptimiitratiisiddhi appears in Anacker 1984 tr., p. 413. The portion that I have translated here reads: "mahayane trai. dhatukam vijiiaptimatrarh vyavasthapyate. cittamatram bho jinaputra yad uta traidhatukam iti siitrat. cittam mano vijiianarh vijiiaptis ceti paryayal;l. cittam atra sasarilprayogam abhipretam. matram ity arthapratisedhartham." 57. Ha111986:14. 58. Ha111986:18. Emphasis mine. 59. Anderson and Smith 1913 ed., p. 147: Ye ca rupiipaga satta ye ca aruppavasino nirodham appajananta, aganwo punabbhavam Ye ca rupe pariiiiiaya arupesu susaIJ\hita nirodhe ye vimuccanti, te jana maccuhayino 60. Anacker (1984 tr., p. 76-77, n. 12) sketches out a brief history of the Buddhist notion of the two and three realms that is essentially like the one I have outlined here., 61. Kochumuttom (1982: 165) makes much of the fact that the word "traidhiitukam" is an adjective rather than a noun and must therefore modify some noun. But, he observes, there is no noun explicitly stated in the sentence, so it should be understood from context

110

Notes

that the noun to be supplied is "cittam." The importance he places ,on this set of observations is that when V asubandhu says the traidhatuka is nothing but vijnapti, he means that the whole realm of subjective experience or thought is made up only of "representations of consciousness," which is his translation of the term "vijfiapti." Hall (1986:22, n.23) feels that Kochumuttom's rendering reduces the entire statement to a tautology saying nothing more interesting than that all subjective experience is subjective. Hall prefers, therefore, to treat "traidhatukam" as a substantival adjective, since, as he correctly points out, words that are formally adjectives can readily stand alone as nouns in Sanskrit without any clearly implied noun to serve as that which the adjective modifies. My own inclination is to side with Hall in this dispute over the grammar of the sentence and to leave the referent of "what is derived from the three elements" deliberately vagu~. 62. Vaidya 1963 ed., p.34: pudgalal). sarhtatil). skandhal). pratyaya lU)avas tatha pradhanam lsvaral). karta cittamatram vikalpyate 63. Many translators treat the term "lokottaram" as referring to something supramundane. Such a translation can be somewhat misleading, for it seems to suggest to some ears that which is supernatural. But all that "/okottara" implies is that which is superior (uttara) to what is found among ordinary folks (loka), or that which is superior to the foolish masses of people (biilaputhujjanii) and their preoccupations with the eight folk characteristics (lokadharma), namely the longing to gain and the dread of being deprived of property, comfort, fame and praise. 64. Literally the expression "vimukti-kiiya" means body of liberation, but the word "kiiya" often appears in iigama literature as that sense faculty by which one experiences happiness and joy, since feelings of intense emotion are often manifested in goose-flesh and shivers. Moreover, in abhidharma literature it is common to use the word "kiiya" to refer collectively to the non-corporeal groups of properties that make up a living being's mental life. 65. Originally a muni was a mendicant who had taken a vow of silence. Silence came to be regarded as the mark of a wise person. The Buddha Gautama, despite his many conversations and talks, was noted for his silence and composure and came to be called by his followers the Great Muni. Given his aversion to the formation of opinions, and given that when one abandons opinions there is much less to say, "muni" was probably an epithet that was deliberately chosen for its conpotations of silence. In this same sentence the word "dharma" has the double sense of property or attribute and of teaching or doctrine. The Sanskrit for these three verses appears in Anacker 1984 tr, p. 423: yadaIambanam vijiianam naivoplabhate tada sthitaIil vijiianamatratve grahyabhave tadagrahat acitto 'nuplambho 'saujnanam lokottaram ca tat asrayasya paravrttir dvidha daus(halyahanitaf:i sa evanasravo dhatur acintyal;t kusalo dhruval;t sukho vimuktikayo 'yam mahamunei}

Chapter 4 Dinnaga's Theory of Knowledge What I shall argue in this chapter is that it is possible to see Diilnaga as a skeptic, very much along the lines of the many Buddhists who preceded him, and that as a skeptic his main interest was not to find a way to increase our knowledge but rather to find a way to subtract from our opinions. To state the matter in a slightly different way, I shall argue that Diilnaga's central interest was to set down the necessary conditions that would have to be met before we could claim that our knowledge had been increased. These conditions, then, can be seen as controls against the proliferation of uncritical and irrational judgements. My hypothesis that Diilnaga's aims essentially were those of the rational skeptic is one that I believe accounts well for. the evidence at hand, but it is one that I have formed only gradually over the course of several years. Rather than inviting anyone to accept the hypothesis at this stage, I shall first simply present the principal features of Diilnaga's system of epistemology as it is developed in two of his texts, the PramlifJasamuccaya and the HetucakranirfJaya, and then I shall offer some comments at the end of this chapter as to why it appears to me that a skeptical interpretation of Diilnaga's epistemology seems suitable.

4.1 Hetucakranir1J.aya

The HetucakranirfJaya is a very brief text in which Diilnaga deals with the problem of confirmation (slidhana). The title of this text can be translated as "Assessment of the cycle of reasons." The word "cakra" literally means circle or wheel, but in more extended usage it means group, set or multitude. It is, therefore, rather like the English word "cycle," which also means group or set in such expressions as "cycle of poems." The word "hetu" can be applied to anything whatsoever that accomplishes a desired purpose. It is generally translated as "cause," and may be

111

112

Diimaga's Theory of Knowledge

understood either as a material cause or a cause of understanding. In the context of logic it may refer either to a property used as evidence for or a sign of some further property, or to a statement that causes another person to believe that something is the case. The HetucakranirfJaya is a systematic assessment of a set of reasons that might be put forward in support of given conclusions along with a very brief account of why each one is or is not a good reason. Before going on to describe how Diimaga approaches that problem, I should make it clear that most other interpreters of the Buddhist epistemologists prefer to treat sadhana as proof rather than as confirmation. But the English word "proof' is one that careful writers usually reserve for such domains as mathematics and logic where theorems that can be demonstrated to follow from axioms are regarded to be conclusively true. But in practical domains and in the empirical sciences, virtually no hypothesis is ever regarded as conclusively true but rather is regarded as consistent with all known evidence but liable to be overthrown in the light of evidence that might be obtained in the indefinite future. Buddhist epistemology is much closer both in its goals and in its methods to the practical reasoning of law and the empirical sciences than to the rigorous disciplines of mathematics and logic, and it is therefore rather inappropriate to use the terminology of these more exact sciences in discussing it. Therefore I shall follow the terminology developed by such modem logicians as Ajdukiewicz, who explains the convention of using the term "hypothesis" to refet to a statement that is subjected to a testing procedure. Ajdukiewicz explains: The testing of a hypothesis H is an operation undertaken to substantiate an answer to the decision question as to whether that hypothesis is true or false. But...hypothesis testing yields a definitive answer to this question only if the answer is in the negative (and that only if we are not inclined to reappraise any of the previously accepted statements that form the body of knowledge K). If the answer is in the affirmative; i.e., if all examined consequences of E prove true, the testing procedure does not provide a definitive answer to the decision question described above, but can merely reinforce the original assumption from which we started when testing H, namely the assumption that H is true. This reinforcement of the original assumption, however, leaves the problem of the truth of the hypothesis in question still open, and it does not conclude its testing. On having found certain consequences which confIrm a hypothesis we remain ready to reject it if another consequence, deduced from it later, proves false. l

In Diilnaga's study of the process of confirmation, the central task is to make a comparison of the relations of each of two classes of individuals with a third class of individuals with the aim of discovering the

Chapter Four

113

relation that the first two classes have to one another. 2 A study of the structure of the HetucakraninJaya suggests that Diimaga saw as his first task that of describing the process by which we can arrive at a judgement of whether one class contains another, whether the two classes are disjoint, or whether the two classes overlap. One class contains another in case every member of the second is a member of the fIrst. Two classes are disjoint in case they have no members in common. And two classes overlap in case there are individuals that belong to both, individuals that belong to the fIrst but not to the second and individuals that belong to the second but not to the first. In Diimaga's writing we find no terms that exactly correspond to the relations of containment, disjunction and overlapping described here. As was the custom among Indian logicians, Diimaga tends to talk not about classes themselves as the extensions of properties but rather of the properties that are the intensions of classes. 3 The relationships he talks about between properties are the pervasion (vyiipti) relation, the incompatibility (virodha) relation and the errancy (vyabhiciira) relation. One property pervades another in case the extension of the fIrst contains the extension of the second. Two properties are incompatible in case their extensions are disjoint. And one property is erratic with respect to a second in case part of the extension of the second is a proper subset of the extension of the first. Since talk about properties can always be translated into talk about classes and vice versa, there is no great harm in discussing Diimaga's logic of properties in the language of the logic of classes that is somewhat more familiar to most readers of English. The first step in arriving at a judgement concerning the relations that obtain between pairs of classes or properties is to examine each individual within a given domain of individuals and to determine for each individual whether or not it is a member of each of the two classes the relation between which one is trying to ascertain. In the Hetucakranin;aya the two classes being compared are the class of individuals that are possessors of the property that is used as evidence (hetu) and the class of individuals that are possessors of the property to be confIrmed (siidhyadharma) through the evidence. The domain of individuals within which these two classes are compared may be seen as a limited sample of the universe of all individuals, the sample being limited in the sense that it excludes all individuals of the class of individuals whose relationship to the property to be confirmed is in doubt. This class of individuals whose relationship to the property to be confmned is in doubt is called the subject (pak$a) of the inference. In the discussions that follow, I shall use the term "induction domain" to refer to this set of observed individuals that excludes individuals in the subject class. In comparing the relationship of

114

Diimaga's Theory of Knowledge

the class H (for hetu-possessors) to the class S (sadhyadharma-possessors) within the induction domain, we find that each individual occupies exactly one of four subdomains or compartments of the induction domain. The four compartments are as follows: 1. 2. 3. 4.

the set of individuals that belong to both Hand S the set of individuals that do not belong to H but do belong to S the set of individuals that belong to H but do not belong to S the set of individuals that belong neither to H nor to S.

In the following set of discussions I shall represent these compartments through a notation adapted from the one employed by John Venn:" (Nl)

HS

(N2)

-HS

(N3) (N4)

H-S -H-S.

In this system of notation the juxtaposition of two or more symbols representing classes indicates the intersection of the classes in question. A tilde (-) before a letter indicates the complement of the class for which the letter stands, the complement of a class being all the individuals in the universe that do not belong to that class. Thus "H"'B" stands for the class of all individuals that possess the evidence but do not possess the property to be confirmed.

Suppose we have the task of assessing the possibility that a given individual has a given property, and suppose that direct examination of the individual in question yields nothing decisive on the question of whether or not the individual that is the subject matter of the inquiry possesses the property to be confirmed. Suppose that what the direct examination does yield is that a second property is present. The question that now arises is: Can the presence of this second property be used as evidence for the presence of the property about which we are in doubt? Looking at the problem theoretically, it can be divided into a number of subproblems. Since a direct examination of the subject matter of our inquiry has yielded nothing conclusive, the investigation must tum to the induction domain, that is, the set of all individuals that are not the subject matter of our investigation. Theoretically speaking, the task is to examine each individual object in the induction domain and to determine for it whether or not it has the property that is being considered as evidence in our original investigation, and whether it has the property to be confirmed in our original investigation. In other words, we must deternilne for each

Chapter Four H-S -H-S 0 + 0 +

(8)

HS -HS 0 + +- 0 0 + 0 + 0 + 0 + 0 + 0 +

9. 10. 11. (12)

+ + + +

+ + + +

+

(13)

0 0 0 0

0 0 0 0

1. 2. 3. (4) 5. 6. 7.

(14) (15) (16)

+

+

0

0 0

+

0 +

0 0 +

0 +

0

115

+ +

0 0 + +

0 0

+

+ +

0

0

Table 4.1: Sixteen Possible Configurations of the Induction Domain

==============================================================

individual which of the four compartments (HS, -HS, H-S, -H-S) it occupies. Once we have examined each individual in the induction domain and detennined which of the four compartments it occupies, we may then move on to the next subproblem, which is to examine each compartment and determine whether any individuals occupy it or whether it remains empty. Practically speaking, of course, what one will do is simply to think of each compartment and try to recall whether any individual in one's experience has occupied it or not. Once it has been determined that a compartment has even one individual in it, it is known that the compartment is occupied, and one can move on to the consideration of the next compartment. The induction domain taken as a whole will have one of sixteen possible configurations. 5 Of those sixteen, Dinnaga considers nine in his HetucakranirtJaya. His method of considering them is to consider pairs of properties that either occur together or do not occur together in individuals. The individuals serve as examples of occupants of the four compartments of the induction domain. The so-called examples given in the HetucakranirlJ.aya are always given in the following order: an individual that occupies HS if any, then an individual that occupies -HS if any, then an individual that occupies H-S if any, and finally an individual that occupies -H-S if any. Thus if four examples are given, it means that all four compartments are occupied (as in CR position nine described below). If fewer than four examples are given, then the first one will always be an instance of an individual that shows that 8 is occupied, and the last will always be an instance of an individual that shows that -8 is occupied.

116

Diflllaga's Theory of Knowledge

The sixteen possible configurations that a domain of 'individuals being examined with respect to two properties Hand S can have are shown in Table 4.1. The numbers enclosed in parentheses, such as "(4)", indicate the configurations that Diilllaga does not consider in his Hetucakranin;aya. In general in the following discussion the fact that a class or compartment C is not empty will be expressed by the notation "C>O", to be read "(The class) C has (more than zero) members." And the fact that a class or compartment is empty will be expressed by "C=O", to be read "C has zero members." In Table 4.1, however, in order to ease reading a nonempty class is indicated by "+" and an empty class by "0". An examination of the seven theoretically possible configurations that DiIinaga does not consider in Hetucakranin;aya shows' that he eliminates just those configurations in which either both HS and -HS are empty (configurations 13, 14, 15, 16) or both H-S and -H-S are empty (4, 8, 12, 16). In other words, he eliminates every case in which the induction domain is either devoid of any individual that belongs to S or is devoid of any individual that belongs to -So As we shall see in greater detail below, the class of individuals of the induction domain that belong to S is called the subjectlike class (sapaksa), and the class of individuals that belong to -S (that is, do not belong to S) is called the unsubjectlike class (asapaksa). Thus it can be seen that Diilllaga considers in the Hetucakranin;aya only those cases in which both the subjectlike class and the unsubjectlike class are occupied by at least one individual. In the Cycle of Reasons itself Diri.naga takes sound or speech (sabda) as the subject of investigation. He considers several properties that sound can be observed to possess, namely, the properties of being knowable, of being created or produced, of being transitory or perishable, of being audible, of being manmade and of being incorporeaL And he considers several further properties that are not directly observable in sound and therefore can be confirmed to exist in sound only through the process of inference using one of its observed properties as a sign. The properties to be confirmed in this way that Diilllaga chooses to investigate are the properties of being eternal, of being transitory, of being manmade, and of being natural (where "natural" is to be understood as coming into existence through some means other than through human effort). The general form of the question to be answered, then is "Given that one has observed property H in sound, can one conclude from that observation that sound also has property S? Or can one conclude that sound lacks property 8? Or can one draw any conclusions at all?" Obviously the question can never be settled by looking within the domain of sound itself, for if one could directly observe a relationship between the evidence H and the property S to be confirmed in sound, there would be no need to appeal to

117

Chapter Four HS

-HS

H-S

-H-S

ETHER

empty

POTTERY

empty

2 S=transitory H= created

POTTERY

empty

empty

ETHER

3 S=manmade H=transitory

POTTERY

empty

LIGHTNING

ETHER

4 S=eternal

empty

ETHER

POTTERY

empty

5 S=eternal H=audible

empty

ETHER

empty

POTTERY

6 S=eternal H=manmade

empty

ETHER

POTTERY

LIGHTNING

7 S=natural

LIGHTNING ETHER

POTTERY

empty

1 S=eternal H=knowabl~

H=created

H=transitory 8 S=transitory H=manmade

POTTERY

9 S=eternal H=incorporeal

ETHER

LIGHTNING empty ATOMS

ACTION

ETHER POTTERY

Table 4.2: Nine Configurations of the Cycle of Reasons

==============================================================

the indirect method of interpreting one property as the sign of another. So the only way that one can establish whether any regular patterns occur in the presence of the two properties H and S in individuals is to look at the universe outside the subject of the inference itself. One must, in other words, look in what I have called the induction domain. The induction domain in all cases considered in the Hetucakranin)aya is, of course, the set of all individuals that are not sounds. The fIrst step undertaken is to see for each of the four compartments mentioned above whether there exists in the induction domain any class that occupies it. The result of this first step may be seen in Table 4.2. If an individual or class of individuals is found in the induction domain that occupies one of the four compartments, its name is recorded. If no individual or class of individuals is found to occupy a given compartment, the absence of an example is indicated by the word "empty."

118

Dinnaga's Theory of Knowledge

Once this tabulation of examples in the form of individuals representative of the four possible compartments of the induction domain has been carried out, our next step is to see whether, on the basis of our knowledge of which of the four compqrtments of the induction domain are empty and which are occupied, we can arrive at any general propositions regarding the relation of the class of individuals that are possessors of the property being used as evidence, which will for the sake of brevity be called H-possessors from here on, and the class of individuals that are possessors of the property to be confirmed, hereinafter called S-possessors. It will be convenient at this point to introduce two further technical terms that are used for variables of classes and two technical terms applied to relations between those classes. It will be remembered that the induction domain is the complement of the subject (pak$a) class and so may be represented by the symbol "-P". The induction domain is itself divided into two mutually exclusive classes, namely the class of all individuals that are S-possessors, and the class of all individuals that are S-non-possessors. The class of S-possessors in the induction domain is traditionally termed the subjectlike class (sapak$a), and it may be represented by the symbol "-PS". The class of S-non-possessors in the induction domain is called the unsubjectlike class (asapak$a) and it may be represented by the symbol "_P-S".6 Special names are also given to two kinds of relation that can occur between the class of H-possessors and either the subjectlike class or the unsubjectlike class. An association (anvaya) relation holds between the evidence and the property to be confirmed just in case there exists at least one individual in the induction domain that is an H-possessor and is also a member of the subjectlike class. But a holds between the evidence and the property to be confrrmed just in case there exists no individual that is a member of both the class of H-possessors and the unsubjectlike class. We may symbolize these two relations as follows: association dissociation It is important to realize that as Dinnaga uses the technical terms "association" and "dissociation," the terms refer to relations that are logically independent. That is, even if one of the relations is known to hold, it is still impossible to infer immediately whether or not the other holds, and if one fails it is equally impossible to infer whether or not the other holds. As we shall see below, post-Dinnaga Indian logicians tended to use the term "association" to name a relation that is identical to that named by "dissociation," and this later usage of the terms came to be

Chapter Four

119

standard for Buddhist logicians beginning with Dhannakirti, a fact that can cause a considerable amount of confusion if the later usage is projected back onto the works of Diimaga. DiIinaga's usage of the terms is more in line with the grammarians' usage, namely to refer to the two relations symbolized above. The technical terms "anvaya" and "vyatireka" as used by Sanskrit grammarians is explained by Cardona as follows: The method of anvaya and vyatireka used to establish the meaningfulness of components and to ascribe individual meanings to components ...consists in observing the concurrent occurrence (anvaya) of a certain meaninf and a certain linguistic unit and the absence of a meaning and a unit.

That is, when a given root or stem or suffix is present, a certain meaning is understood, but when that root or stem or suffix is taken away or replaced by another, then the original meaning is no longer understood or another meaning arises in the mind of the hearer. Here in the usage of the grammarians, anvaya and vyatireka are logically independent rather than being contrapositive propositions. Let us now consider the problem of how we can move from knowledge of compartments within the induction domain to the formulation of a universal affirmative proposition regarding H-possessors and S-possessors. Before going any further I must make it clear that as I use the term "universal affirmative proposition" in the following discussions I mean a proposition of the form "Every A is B" in which it is understood that the subject term A names a class that is not empty. In this respect it differs from a proposition that states a dissociation relation as described above. The propositional form that best captures the state of affairs in which a dissoci.ation relation holds between classes A and B is the obverse proposition derived from the universal affirmative. An obverse proposition is the result of an immediate "inference from a given proposition of one in which the quality of the original proposition is changed, and at the same time the 'sign' of its predicate reversed. "8 So the obversion of "Every A is B" is "No A is non-B." I take it that when the class A has no members,. this obverted proposition is true, for it simply says that there exists no member of the class A that is non-B, a state of affairs that obtains automatically if there exists no member of the class A. Thus if the class A has members, then "Every A is B" and "No A is non-B" are IQgically equivalent. But if class A has no members, then I shall regard "No A is B" and "No A is non-B" as true, while I shall take "Every A is B" to be indeterminate in truth-value. 9

120

Diimaga's Theory of Knowledge

Dinnaga's views on the conditions necessary to warrant asserting a universal proposition apparently rest upon the intuition that we are not likely to believe that one property always occurs with a second property unless the two following conditions obtain: 1. there is some precedent of the two properties actually occurring together, and 2. there is no reason to deny that the first always occurs with the second. The association (anvaya) relation between the class of H-possessors and the class of S-possessors serves to justify our entertaining the possibility that every (member of class) H is (a member of) S, although it is of course not a sufficient ground for asserting that every H is S. If, however, there is also a dissociation (vyatireka) relation between Hand S, then there is no reason to deny that every H is S. If the association relation fails to hold while the dissociation relation holds, then we have no warrant to assert a universal affirmative proposition. The instance that Dinnaga gives of association failing while dissociation holds is the case in which the property of being audible is under review as a tentative sign of any other property in sound. If the subject of the inference is the class of sounds, the induction domain is of course the class of non-sounds. But the property of being audible fails to occur at· all in this induction domain, for only sounds are audible. Hence the association relation between audibility and any other property fails to hold. Therefore, we have no justification either to believe that every H is S or to deny that no H is S. But the dissociation relation does hold in this case, because it is not the case that some audible non-sound is non-S, regardless of what the property S may be. But if there is no reason either to affirm or deny either the universal affirmative or its contrary, we must simply suspend judgement altogether on the relation between audibility and other properties. Let us now return our attention to examples given in the H etucakranin;aya. It will be observed that for each of the nine positions in the Cycle the presence or absence of an example allows us to assign a truthvalue to each of four particular propositions of the following forms:

1. 2. 3. 4.

some H is S some non-H is S some H is non-S some non-H is non-So

Chapter Four CATEGORY Ths ANVAYA Th-s I II ill

IV

T T F F

HOLDS HOLDS' FAILS FAILS

F T T F NIhs

121

VYATIREKA NIhs Nlli-s Ehs HOLDS FAILS FAILS HOLDS

F F T T

T F F T

Eh-s CR POSmON

F F T

T F F

2,8

?

7

5

1,3,7,9 4,6

=Ehs; NIh-s =Eh-s; Ehs =Ah-s; Ehs =Ahs

Table 4.3: Four Categories of Association-Dissociation (anvaya-vyatireka) Relation

If we now consider just the values assigned to propositions of the form "Some H is .... " and disregard those of the form "Some non-H is ... ," we find that the nine Cycle of Reasons (CR) positions can be divided into four distinct categories according as the association and dissociation relations hold or fail to hold. This division can be seen in Table 4.3. In Table 4.3 "Ihs" reads "some H is S." "Ehs" reads "no H is S." "Ahs" reads every H is S." "N... " reads "It is not the case that...." The tilde (-) before a lower case letter signifies the negation of that term.

The results of Diilnaga's reflections on the nature of inference were expressed in a set of statements describing the three characteristics (trirupa) of legitimate evidence, about which more will be said later in this chapter. The results of the examination that is shown in Table 4.3 were expressed specifically by the second and third statements in that set. In these two statements Diilnaga states in effect that in order to derive a universal proposition of the form "every H-possessor is an S-possessor," it must be observed that

1. there are H-possessors in the domain, and 2. there are no H-possessors that are S-non-possessors. The first characteristic of legitimate evidence described in Diilnaga's formulation specifies that the property used as evidence must belong to the class to which the subject of the inference belongs. Presumably what is intended is that the property used as evidence belongs to every member of the class to which the subject belongs, but it must be noted that Diilnaga does not always carefully distinguish between inferences in which the subject is an individual and inferences in which the subject is a class of individuals. In any event when it is known that the subject of the inference is not an empty class and that no member of the subject class is an H-nonpossessor, then the first characteristic of legitimate evidence is in place. This characteristic is called "pa~adhannatii," or the fact of being a property of the subject. Using our notation we may symbolize the three

122

Dinnaga's Theory of Knowledge

characteristics of legitimate evidence as described by Diimaga as follows: Legitimate evidence has the characteristics of 1. PH>O & P-H=O 2. -PHS>O 3. -PH-S=O

being a property of the subject association dissociation

Supposing we have now conducted our investigation of individuals in the induction domain and of individuals in the domain of the subject, and supposing also that our investigations have shown that every Hpossessor in the induction domain is an S-possessor and every member of the domain of the subject is an H-possessor, we can now ask whether all this knowledge is sufficient to compel us to conclude that every member of the subject domain is an S-possessor. To this question it is clear that we must give a negative reply. To see why this is so, consider Table 4.4. In this table it can be seen that when three properties are being considered with reference to one another, the universe is divided into 23 = 8 compartments. The conventions that I shall use to refer to these compartments are shown in the Venn diagram on the left. In the Venn diagram on the right, some of the compartments are shown shaded, representing empty compartments. The shading of compartment 1 indicates that -PH-S=O, which means that there is dissociation of the evidence from the absence of the property to be confirmed. Compartment 2 is not shaded, indicating that -PHS>O. Compartment 4 is not shaded, indicating that PH-S>O and therefore that PH>O, and compartments 6 and 7 are both shaded, which means that P-HS=O and P-H-S=O and therefore P-H=O; taking compartments 4, 6 and 7 all together allows us to see that the characteristic of the evidence being a property of the subject is in place. What the diagram taken as a whole indicates, then, is a state of affairs in which all three of the characteristics of legitimate evidence that Dinnaga described obtain. Compartments 4, 5, 6 and 7 together malce up the circle that represents the subject domain, and what the diagram indicates is that any compartment within that circle that is not shaded is a compartment to which an individual in the subject domain may belong. Compartment 4 is not shaded. Therefore, it can be seen that even when all three of the characteristics of legitimate evidence as described by Dinnaga are in place, we can still not be sure that the proposition "every P is S" is true. The question that now arises is whether this spells failure for Dinnaga's attempt to described the conditions of legitimate evidence. Before trying to answer that question, let us examine the Cycle of Reasons in greater detail.

Chapter Four

123

8 Legend for eight compartments of left diagram: 1 =-PH-So 2 =-PHS. 3 = -P-HS. 4 = PH-So 5 = PHS. 6 = P-HS. 7 = P-H-S. 8 = -P-H-S. Compartments 1, 2, 3 and 8 represent the induction domain. Compartments 4, 5, 6 and 7 represent the subject domain. Legend for right diagram: the shaded areas represent empty compartments. Table 4.4: Eight compartments of a three-property domain ==============================================================

As we saw above in Table 4.3, the nine types of situation considered in the Cycle of Reasons can be divided into four categories according as the characteristics of association and dissociation are found in the evidence. Let us now examine these four categories in greater detail. Category I contains two situations, the second and eighth considered in the Cycle of Reasons, in which the association and dissociation relations both hold. The situations as Diimaga describes them are as follows: (CR2) Sound is transitory. Because it is created. Consider pottery [created and transitory] and ether [not created and not transitory]. Present in the subjectlike class [that is, in all transitory things in the induction domain] and absent in the unsubjectlike class [that is, in nontransitory things in the induction domain], the evidence [that is, the property be being created] is good. (CR8) Sound is transitory. Because it is manmade. Consider pottery [manmade and transitory] and lightning [not manmade but transitory], and consider ether [not manmade and not transitory]. Both [present and absent]

124

Difmaga's Theory of Knowledge

ether

ether

Upper left diagram shows CR2 in which the three characteristics of evidence are in place. Upper right shows CR8 in which three characteristics are in place. Lower left shows CR2 with added assumption that if dissociation holds in the induction domain, it also holds in the subject domain. Lower right shows CR8 with same assumption. Table 4.5: Cycle of Reasons situations 2 and 8 ==============================================================

in the subjectlike class [that is, present in some transitory things and absent in other transitory things in the induction domain] and absent in unsubjectlike cases, the evidence is good. A look at the diagrammatic representation (Table 4.5) of these two situations shows that the evidence will be successful only if we make the

Chapter Four

125

following assumption: if the dissociation relation holds in the induction domain, then it also holds in the subject domain. That is, we must assume that if -PH-S=O· then PH-S=O. The implications of making this assumption will be explored below. Category II of the Cycle of Reasons contains four situations, the first, second, seventh and ninth considered in the Hetucakranin;aya, in which association holds and dissociation fails. The four situations as described by Dirinaga are as follows: (CRI) Sound is eternal. Because it is knowable. Consider ether [knowable and eternal] and consider pottery [lmowable and not eternal]. Present in all subjectlike and present in all unsubjectlike cases, the evidence is ambiguous. (CR3) Sound is manmade. Because it is transitory. Consider pottery [transitory and manmade], and consider lightning [transitory and not manmade] and ether [not transitory and not manmade]. Present in all subjectlike cases and both [present and absent] in unsubjectlike cases, the evidence is ambiguous. (CR7) Sound is natural. Because it is transitory. Consider lightning [transitory and natural] and ether [not transitory but natural], and consider pottery [transitory but not natural]. Both [present and absent] unsubjectlike cases and present in all unsubjectlike cases, the evidence is ambiguous. (CR9) Sound is eternal. Because it is incorporeal. Consider ether [incorporeal and eternal] and an atom [not incorporeal but eternal], and consider action [incorporeal and not eternal] and pottery [not incorporeal and not eternal]. Both [present and absent] in subjectlike cases and both [present and absent] in unsubjectlike cases, the evidence is ambiguous. A look at the diagrams for these situations (Table 4.6) shows that in none of these four cases are the compartments PHS and -PH-S empty. But if neither of these compartments is empty, then we have no justification to assume that their compartments outside the induction domain, that is, PHS and PH-S respectively, are empty either. In fact, if the same relations hold in the subject domain as in the induction domain between the classes H

126

Dirinaga's Theory of Knowledge

ether

poUery

Upper left shows CR1, upper right CR3, lower left CR7 and lower right CR9. Table 4.6: Cycle of Reasons situations 1,3, 7 and 9 ==============================================================

and S, it is natural to assume that some members of the subject class will be S-possessors and some will be S-non-possessors. Thus in all four cases the evidence under consideration fails to warrant the conclusion that every Pis s. Category III of the Cycle of Reasons contains two situations in which both the association relation and the dissociation relation fail to hold in the induction domain. Dirinaga presents these arguments as follows:

Chapter Four

127

(CR4) Sound is eternal. Because it is created. Consider ether [not . created but eternal], and consider pottery [created but not eternal]. Absent in subjectlike cases and present in all unsubjectlike cases, the evidence is incompatible [with the property to be conflrmed]. (CR6) Sound is eternal. Because it is manmade. Consider ether [not manmade but eternal], and consider pottery [manmade but not eternal] and lightning [not manmade and not eternal]. Absent in subjectlike cases and both [present and absent] in unsubjectlike cases, the evidence is incompatible. As the diagrams for these arguments show (Table 4.7), the evidence will succeed in confirming the absence of the property to be confmned only if we make the following assumption: if the association relation fails in the induction domain, then it fails in the subject domain; that is, if "'PHS=O thenPHS=O. Category IV contains only the fifth situation considered in the Hetucakrani7lJaya. Diimaga presents this argument as follows: (CR5) Sound is eternal. Because it is audible. Consider ether [not audible but eternal], and consider pottery [not audible and not eternal]. Absent in subjectlike cases and absent in unsubjectlike cases, the evidence is unique [to the subject domain]. We saw above in studying the situations of categories I and ill that we were to make the assumption that whichever compartment of the class of H-possessors is empty in the induction domain, its counterpart in the subject domain is also empty. That is, if "'PHS=O, then we must assume that PHS=O, and if "'PH"'S=O, then we must assume that PH-S=O. Thus it might seem reasonable prima facie to make a similar assumption for CR5, in which case we might assume that both PHS and PH"'S are empty. The result of following out this assumption, however, would be to render the subject domain entirely empty. But if the subject domain is entirely empty, then the flrst characteristic of legitimate evidence, that is being a property of the ,subject, cannot be in place for the simple reason that there is nothing in the subject domain. Thus if the fIrst characteristic of legitimate evidence is in place, we cannot assume that both PHS and PH"'S are empty. It is possible, given that being a property of the subject and the dissociation relation obtain while association fails, that either both PHS and PH"'S are

128

Dinnaga's Theory of Know ledge

lichlnlng

Upper left diagram shows CR4 in which the association and dissociation relations fail to hold. Upper right shows CR6 in which the association and dissociation relations fail to hold. Lower left shows CR4 with added assumption that if dissociation holds in the induction domain, it also holds in the subject domain. Lower right shows CR6 with same assumption. Table 4.7: Cycle of Reasons situations 4 and 6 ==============================================================

occupied or that only one of them is occupied; it is not possible on the given evidence to conclude anything more than that at least one of the two compartments must be occupied. And so Dinnaga's conclusion is that arguments appealing to evidence having characteristics such as those described in CR5 are inconclusive, because the evidence itself is

Chapter Four

129

/lUC/cry

Table 4.8: Cycle of Reasons Situation 5

==============================================================

ambiguous in the same way that the types of evidence discussed in CR1, CR3, CR7 and CR9 are ambiguous. Diimaga's Hetucakranin)aya is a very spare presentation, consisting only of nine situations and an indication of whether the evidence is good or not for confIrming the property to be confIrmed by it, as well as a very terse account of what makes the evidence good or bad. There is no explicit account of what procedure one is to go through in determining whether a line of reasoning is good. Nevertheless, one can reconstruct a defInite procedure, a series of steps, that one might go through in order to arrive at the results presented in the Cycle of Reasons. It seems fairly clear that in assessing the success of a line of reasoning that arrives at the conclusion that all members of the subject domain possess a given hypothetical property on the grounds that all members of the subject domain possess a given sign, one is to go through the following steps in the order indicated below: 1. Examine the members of the subject domain and ascertain whether the evidence is present in each and every member. If it turns out that PH>O and P-H=O, go on to the second step. If not, declare the evidence inadequate to yield the hypothetical conclusion.

130

Diimaga's Theory of Knowledge 2. Examine the members of the induction domain and ascertain whether the association relation (-PHS>O) holds and whether the dissociation relation (-PH-S=O) holds. If both hold or if both fail, go on to the third step. If only one holds, declare the evidence inadequate to yield the hypothetical conclusion. '-\,--

3. Assume that the counterpart of whichever compartment of the class of possessors of the evidence is empty in the induction domain is also empty in the subject domain.

By going through the three steps outlined above, we can write out a sequence of premisses and conclusion of the kind of argument endorsed by Diimaga in Hetucakranin;taya. Suppose we have carried out the required investigations of individuals in the subject domain and in the induction domain and have found that all three characteristics of legitimate evidence are in place. We are then in a position to record the results of these investigations and the conclusions drawn from them in the following seven steps: 1. PH>O

2. P-H=O 3. -PHS>O 4. -PH-S=O

5. PH-S=O 6. H-S=O 7. P-S=O

by observation being a property of the subject association, by observation dissociation irlductive assumption warranted by 3 and 4 derived from 4 and 5 derived from 2 and 6

It will be observed that neither the association (anvaya) relation (step 3) nor the dissociation (vyatireka) relation (step 4) plays a direct role in arriving at the conclusion (step 7). Rather, their function is to work together to warrant the inductive assumption (step 5) that when conjoined with the dissociation relation yields the premiss (step 6) that allows us to derive the conclusion. It will also be observed that none of the above seven steps corresponds exactly to the major premiss, minor premiss and conclusion of the Barbara syllogism known to European logic. We can, however, easily derive the components of a Barbara· from these seven steps. A Barbara comprises three universal affirmative propositions of the form "every A is B," which it will be recalled is to be interpreted as saying "There are A's and no A is non-B." The major premiss of the Barbara reads, using Indian terminology, "Every H is S." This can be derived from step 1, which says "There are individuals in the subject domain that are Hpossessors," and from step 6, which says "No H-possessor is an S-nonpossessor." Similarly, we can derive the Barbara minor premiss, "Every P is H" from steps 1 and 2. And we can derive the conclusion""Every P is S" from steps 1 and 7.

Chapter Four

131

After going through these seven steps, one will be in a position to make a reasonable judgement that every member of the subject domain is a possessor of the property to be confirmed. But the judgement, reasonable though it may be, is still far from secure, since it rests upon an assumption that patterns observed in the induction domain are uniform throughout the universe. The difficulties involved in making this assumption will be discussed in greater detail below. The Hetucakranin)aya offers little more than a sketch of Dinnaga's theory of inference. It is an important text in that the examples it offers of various situations became the stock examples for most later Buddhist epistemologists and also in that it contains in germinal form the key ideas that later more expansive works on inference tried to articulate with greater precision. There are problems in its interpretation, but we shall return to these after examining Dinnaga's most extended work on epistemology.

4.2 The PramiilJasamuccaya The text in which Dinnaga's work is presented in its most mature form is his Pramtif)asamuccaya, a title that he gave it since it is for the most part a collection (samuccaya) of ideas that he had presented earlier in his career in various smaller works dealing with the problem of the limits of knowledge. While it is true that much of the contents of this work had appeared in earlier works, the presentation of his ideas tends to be more orderly and rigorous here than in earlier works, and we also find ideas presented here that had not been articulated in works composed in his younger days. The book comprises six chapters, the contents of which are arranged as follows. The first chapter introduces the general problem of the means of acquiring new knowledge, stating that there are essentially two mutually opposed aspects of things that can enter our awareness, namely a particular aspect that, being a physical feature of the world existing outside the mind, can be cognized only through the physical sense-faculties, and a general aspect that, being conceptual in nature, can be cognized only through the intellect. Incidentally, Dinnaga is traditionally presented as advocating the position of the Yogacara school of Buddhism, which in turn is traditionally presented as denying the reality of objects external to awareness. As I have already argued in section 3.3.2, the Yogacara school put forth a position closer to phenomenalism than to subjective idealism, and as I shall argue

132

Diilnaga's Theory of Knowledge

in chapter five Diilnaga's position in texts such as the Alambanaparfk.ya is also a form of phenomenalism. But in the Prama1j,asamuccaya, in contrast to his works in which he explicitly argues for phenomenalism, he shows every indication of having intended his work to be acceptable to both those who denied and those who affIrmed that we can know external objects directly. Thus for the phenomenalist, Diilnaga's use of such terms as "external object (bahyartha)" can be regarded as a concession to the conventional manner of speaking. lO The form of awareness that consists in the acquisition of information about particulars is called sensation (pratyak.ya) and is the topic of the flrst chapter of Prama1j,asamuccaya. Diilnaga first presents his own views on sensation, then he criticizes the views on sensation of his forerunner Vasubandhu, and of the Nyaya school, the Vaise$ika school, the Sfui:lkhya school and the Mlmamsa system. An English translation of this first chapter has been published along with an informative introduction and a very thorough set of notes by Hattori Masaaki (1968). The question of how we can acquire information about objects not within the range of our physical senses, and the nature of that information, is taken up for discussion in the second chapter of Prama1j,asamuccaya. As with the flrst chapter, the second is divided into a presentation of Diilnaga's own views followed by a criticism of alternative views. A Japanese translation of the part dealing with Diilnaga's own views has been published by Kitagawa Hidenori (1965:73-125), and an English translation of this part appears in chapter six of the present book. The topic of the third chapter is how knowledge that we have acquired ourselves can be imparted to others. This chapter deals in particular with the proper presentation of argument in formal debate. A Japanese translation of the flrst section of this chapter has also been published by Kitagawa (1965:126-238.) The fourth chapter deals with the role of the example in the presentation of arguments in formal debate, and the sixth chapter deals with the refutation of the opponent's position and errors that if committed render refutation invalid. These two chapters were translated into Japanese by Kitagawa (1968:139-281, 282-351). Chapter five of Prama1j,asamuccaya treats a variety of topics connected with the relation between language and that which is communicated through it. The essential point of this chapter is to show that language conveys knowledge in the same way and of the same nature as that which is conveyed by an inferential sign (liizga), and that therefore cognition based upon verbal communication is essentially" the same as inferential cognition. There is also a considerable amount of discussion of

Chapter Four

133

the nature of verbal appOSitIOn and the qualification of one word by another, on which views Diilllaga presents his own views and criticizes alternative theories. 'A translation of this chapter appears in chapter seven of the present book. We can see in the above brief summary of the topics dealt with and the arrangement of those topics in the Pramtir;asamuccaya two distinct features of Diilllaga' s thought, features that were picked up and developed by subsequent generations of Buddhist logicians. The first of these is the differentiation of sensation from inference on the basis of the kinds of object cognized by them, and the second is Diilllaga's clear differentiation between reasoning (svtirthtinumtina) as a process of acquiring new information and demonstration (partirthtinumtina) as a process of presenting information to others in order to persuade them. This distinction had not always been clearly made before Diilllaga's time, although Frauwallner has shown that Dinna!ja probably did borrow this important distinction from Sfuhkhya logicians. I

4.2.1 Sensation in the PramalJasamuccaya It has been mentioned above that Diilnaga's point of departure in the flrst chapter of the Pramtir;asamuccaya is to draw a radical distinction between two kinds of thing that can be the object of awareness. He says in the second verse of the first chapter and its commentary: Sensation and reasoning are the only two means of acquiring knowledge, because two attributes are knowable; there is no knowable object other than the peculiar and the general attribute. I shall show that sensation has the peculiar attribute as its sub~ect matter, while reasoning has the general attribute as its subject matter. 2

As this passage shows, it is Dinnaga's contention that one type of knowable item is that which is immediately present to one of the senses, while the other type of knowable is that which is not present to the senses but which nevertheless enters into our awareness. According to Diimaga, the moment we begin to take the peculiar attributes that are the data of the separate sense faculties and synthesize them into multi-propertied "objects" or "individuals" or to identify those peculiar attributes as individual instances of some class, we are engaged in cognitive activity of a sort different from pure sensation. Dinnaga specifies that where sensation ends and judgement begins is in the association of a thing with a name, a genus, a quality or an

134

Diimaga's Theory of Knowledge

accidental attribute. He spells this out in Pramii1]asamuccaya 1:3cd and its commentary. Sensation is devoid of structure. That cognition in which there is no structure is sensation. What kind of thing is this so-called structure? Attaching a name, a universal and so forth. For in the case of proper names, a thing specified by a name is expressed: "I/ittha"; in the case of general terms a thing specified by a universal is expressed: "cow"; in the case of adjectives a thing specified bya quality is expressed: "white"; in the case of verbs [and deverbatives] a thing specified by an action is expressed: "cook"; in the case of substantives a thing specified by a substance is expressed: "staffed" or "homed." Concerning this, some say a thing specified by a relation is expressed. But others say a thing specified by words that are devoid of meaning is expressed. That in which this structure is not present is sensationP

For Diimaga the term "sensation (pratyalqa)," which is used to refer either to the cognitive process or to the object cognized, is very similar to the notion of "sensing" and "sensum" (or "sense datum") as used by such twentieth century thinkers as C.D. Broad, who speaks of sensa as follows: We may generalize the theory of sensible appearance as follows: Whenever I truly judge that x appears to me to have the sensible quality q, what happens is that I am directly aware of a certain object, y, which (a) really does have the quality q, and b) stands in some peculiarly intimate relation, yet to be determined, to x....Such objects as y I am going to callSensa. 14

In a further elaboration of his notion Broad writes: Under certain conditions I have states of mind called sensations. These sensations have objects, which are always concrete particular existents, like coloured or hot patches, noises, smells, etc. Such objects are called sensa. Sensa have properties, such as shape, size, hardness, colour, loudness, coldness, and so on. The existence of such sensa, and their presence to our minds in sensation, lead us to judge that a physical object exists and is present to our sense<;. To this physical object we ascribe various properties. These properties are not in general identical with those of the sensum which is before our minds at the moment. ...The sensa that are connected with a physical object x in a certain speciaIl y intimate way are called the appearances of that object to those observers who have these sense data. The properties which x is said to appear 10 have are the properties which those sensa that are x' s appearances really do have. 15

The sensum, according to sense-datum theorists, is that aspect of our experience that is certain and beyond question and quite direct in the sense of involving no interpretation or understanding or structure (kalpanii). Diilnaga's use of the term "pratya~a" is restricted to experience and

Chapter Four

135

experieptia very much like the sensation and sensa of Broad and others. He would, I think, find 110 problem in endorsing a view such as that expressed by Hospers in the following passage: Sensing is different from perceiving. We sense sense-data; we perceive· physical objects. Perception is impossible without sensing (without something given to sense), but it involves more. When we open our eyes we have certain visual experiences--sense-data; in this we are passive, and cannot help what we see. But in addition to this passive intake of sense-data· there occurs an activity that we may call interpretation ....We are classifying our present experience into molds already established by previous experiences. 16

It is just this process of interpretation, organizing data according to

classifications provided by past experiences, without which process there is no understanding and no use of language, that Diimaga seems to have in mind when he speaks of kalpana, a word that literally means putting into order, arranging, forming or structuring. According to Dinnaga, we take the raw data of sensation and attach to them some name, which amounts to assigning them to classes, and when we do so we are no longer dealing just with what is at hand but with a shared something, a general attribute that belongs not only to the things immediately present to the senses but also to objects remembered from the past or anticipated in the future or in some other way not present to the senses. And so, on the grounds that thinking or judgement or reasoning (anumiina) is a complex cognitive act having as its subject matter this shared or general attribute (samanyalalqafJa) of what is not present to the senses, Dinnaga regards it ·as a cognitive process of a sort that is essentially different from sensation, which is a simple cognitive act dealing only with what is immediately present. Further light may be shed on how Dinnaga distinguished sensation from judgement by a review of which kinds of objects of awareness he explicitly said could and which he said could not be regarded as sensa. It has already been noted that the objects in the fields of operation of the five physical sense faculties are regarded by Dinnaga as sensa. But in addition to these five physical sense faculties, Dinnaga, as was the custom in contemporary Indian theories of psychology of cognition, also acknowledged a sixth sense faculty, namely the mind (manas); The objects in the mind's field of operation ate, according to this view, mental events of all types. Thus all mental events, even those that are not sensations, are sensa. Accordingly, Dinnaga classifies all acts of awareness themselves as sensa, because an act of awareness is known directly even if the object of that awareness is not. 17 And finally the object of any act of awareness that is entirely free of the preconceptions arising from previous experience or education is regarded as a sensum. In this connection Dmnaga says

136

Diimaga's Theory of Knowledge Meditation adepts observe just the object itself with an observation that is entirely free of preconceptions arising from the instructions of teachers. The observation of meditation adepts of just the object itself, which observation is unspqiled by preconceptions that come down from tradition, is also sensation. IS

In this passage Diimaga shows an appreciation of the psychological fact

that doctrine, and all the other training we get from the teachers who pass traditional wisdom down to us, serves to flavour and even distort our view of the world. Teaching predisposes us to see things in particular ways and even to select the things we see at all. It takes a great deal of effort to overcome the biases one learns from a teacher, but when these biases are overcome the result is an ability to see things directly and clearly without reflection and reaction. This ability is what Dirinaga calls sensation of a meditation adept (yogipratyak~a). That there is an intimate connection between Dirinaga's discussion of sensation and the question of certainty is suggested by a number of considerations that it would do well to examine here. At least one of the reasons that one might regard acts of awareness as sensa is that we are perfectly safe in saying that the fact of awareness itself cannot be denied. But more than that, we are also safe in saying for any single act of awareness that jt is the cognitive act that it is and no other. Each cognitive act, in other words, has an identity or an identifying element. What it is that constitutes the identity of a given act of awareness is just its content, that is, a particular combination of shapes and colours or a particular odour or a particular sensation of pain or pressure or a particular sense of longing or disgust or even a particular thought such as "Tomorrow is Friday." Now it would be absurd to say of the thought "Tomorrow is Friday" that the proposition "Tomorrow is Friday" only seems to be its content but that its reaL content is something else of which the person having the thought is unaware or about which he is mistaken. It may be that "Tomorrow is Friday" is a false proposition at the time that it constitutes the content of a thought, but it is impossible 19 be in error regarding its being the content of the thought of which it seems to be the content. Similarly, if one has an awareness of blue, blue is certainly the content of that particular awareness, even if there is in fact nothing blue outside the cognition for one to be aware of. Hence shapes, odours, pains and so on are counted, along with the cognitive acts of which they are the contents, as sensa. It is important to note that sensa are taken as the subject matter of sensation not as necessary factors of the world but only as necessary factors of awareness. That this is so is clear when one considei's what Dirinaga regards not to be among the objects of sensation. At Pramii1J.a-

Chapter Four

137

samuccaya 1:7-8 he specifically mentions a number of kinds of awareness

that should not be regarded as sensation. Ruled out as acts of sensation are all erroneous cognitions, because they are complex cognitive acts involving the superimposition of mental constructions and interpretations upon sensa. Similarly, all cognitions that involve conventions are ruled out as sensations; the conventions intended here are those of speech such as are shared by a linguistic community as well as those conventions of personal habit that individuals form as a result of their particular experiences. 19 Even though such superimpositions of conventions upon experience may be sanctioned by society and may be consistent with the rest of our experience, and are in a sense therefore correct or true thoughts, they cannot be regarded as instances of coming directly into contact with just things themselves in the way that sensation does. One further mental activity that Dinnaga rules out as sensation is the formation of such attitudes as desire or aversion towards objects. The formation of attitudes is an act of judgement rather than an act of sensation, for feeling desire or aversion involves superimposing upon a sensum something the sensum itself does not intrinsically have, namely, value or repulsiveness. It is allegedly the view of the unreflective masses of ordinary people that we desire things because those things have an intrinsic value or worthiness, and we avoid things that have an intrinsic loathsomeness. But the view of the Buddhist is typically that things in themselves are empty of an intrinsic capacity to evoke emotional reactions; instead, these emotional reactions consist in patterns of behaviour that we acquire through habit and unreflective acceptance of social conventions. It is not that we strive for a goal because the goal is worthwhile, but rather that the goal is worthwhile because we strive after it. By altering these patterns of behaviour, the individual can gradually alter the quality of emotional reactions that are habitually experienced as a part of observing an object. This is why serious Buddhist training required the individual's removal from family and friends and other social contacts that were responsible for shaping the emotional habits that stand in the way of true peace of mind. By living in a regularized environment in which overt striving after goals is a type of behaviour that is actively discouraged, and in which accepting whatever one gets without complaint is encouraged, the monk has an opportunity that people in the world very rarely have to cultivate the habit of seeing things just as they are in themselves. Finally, Dinnaga argues that while the visual field, the audible field and so forth are sensa, when these data are attributed by the mind as various characteristics belonging to a subject or a property-bearer (dharmin), this act of attribution is not an act of sensation, nor is the

l38

Diimaga's Theory of Knowledge

property-bearer a sensum. He writes "no knowledge at all of a possessor of properties that has many characteristics is derived from a sense faculty. ,,20 Thus according to Diimaga, there can never be a direct awareness of a macroscopic object such as an apple; rather, one is aware of a set of radically distinct properties such as a patch of green colour, a sensation of hardness experienced by the fingers, the sensation of heaviness, a particular taste, and a distinctive aroma, all of which are then associated with one another and synthesized into a single object of perception, perception being the mental act of interpreting disparate _sense data and simplifying them into units that are manageable by the consciousness. Why does Diimaga insist that shapes and tastes are objects of sensation and as such are the givens out of which we derive a notion of material objects outside awareness while the material objects themselves are never sensa? Diimaga offers no explicit explanation, but I am inclined to venture the hypothesis that his motivation rests on a consideration of the limits of certainty. The property that is unmistakably the content of awareness is not unmistakably the property of an object outside our awareness; we can be certain that a given colour is a feature of experience, but it is possible to be wrong in thinking that it is a feature of the world. A person, for example, who has a visual experience of red spots can be certain that red spots and nothing else are the content of his experience, and he can even be certain that those very red spots and no other red spots are the red spots he is experiencing, for it would be absurd to suggest otherwise. But it is not absurd to suggest that the contents of an experience of red spots have no counterpart in the world outside experience. This, as we saw in the discussion of Vasubandhu's phenomenalism, is the point of the arguments advanced in the Vijnaptimatratasiddhi. Indeed, it is not entirely absurd to suppose that none of our experiences have exact counterparts in the world outside experience; it may not be a comforting idea, but it is certainly quite a plausible idea that all of our experience is a distortion of the world as it actually is but that the distortions are not usually so serious that they altogether prevent our being able to function with some success. One can make an even stronger claim and say that it is not absurd--not, in other words, necessarily false--to suggest that there simply is no world outside experience. Saying that there is an external world carries with it a greater risk of being wrong than saying that I now see red spots before my eyes. Thus the hypothesis that there exist objects outside cognition that are the basis of our experiences is one the truth of which must be determined by other means than through sensation itself. If it should somehow tum out to be certain that external objects exist, then it would be certain for the reason that a judgement is certain and-not for the

Chapter Four

139

reason that sensation is certain. The reason that a sensation is certain is that its being erroneous is. out of the question or unthinkable. Evidence for the claim that Dinnaga restricts the term "sensation (pratyaksa)" to cognitions that are certain in the sense of error being impossible is to be found in his own commentary to PramiilJasamuccaya 1:17, where he discusses the following definition of the term "pratyaksa" that is given in the Nyayasiitra 1.1.4., which says "Pratyaksa is a cognition

arising from a sense faculty's contacting an object, which cognition is inexpressible, inerrant and definite by nature. ,,21 In his criticism of this definition Dinnaga's point is that no qualifying expressions should be introduced into a definition unless they add further refinement or clarification into what is being talked about. But, he argues, two of the qualifying expressions in the Nyayasiitra definition are redundant, namely, the expression "inexpressible (avyapadefya)" and the expression "inerrant (avyabhiciirin)." The reason they are redundant is that there is no cognition arising from a contact of a sense faculty with an object except that which is inerrant and inexpressible. Dinnaga's point is that to qualify pratyaksa as inerrant is to suggest that it is being distinguished from cognition arising from contact of a sense faculty with an object that is erroneous, much as the phrase "positive prime number" suggf<sts that the qualifier "positive" is being used to eliminate from discussion prime numbers that are negative. But of course there are no negative prime numbers, hence one need not take pains to eliminate them by using the word "positive." Similarly, cognition born of faculty-object contact is necessarily non-erroneous in Dinnaga's view, since a sensation cannot "stray" (vi-abhi-CAR) from its sensum, and so no effort need be made to eliminate what is in any event impossible. Now if sensation is by its very nature inerrant, the question arises as to how we are to classify misperceptions of the sort that arise when the senses are tricked. Dinnaga's interpreter DharmakIrti argued that errors can occur at the level of sensing, such as when we are riding in a moving vehicle and sense that we are stationary and the landscape is moving. But in the work of Dinnaga himself the position that is taken is that what is erroneous is not the sensation itself but the further thinking we do about what is sensed. What is sensed is motion. There can be no error about that. Where there can be error is in our arriving on the basis of this sensation at the conclusion "something is in motion," or even more specifically "the shoreline that I am watching from the boat on which I am sitting is moving." As Dinriaga puts it, "it is impossible too for the object of awareness itself to be errant, for errancy is only the content of misinterpretation by the mind. ,,22 It has already been pointed out by Hattori

140

Diimaga's Theory of Knowledge

that Drnnaga, unlike Dharmaldrti and later Buddhist epistefnologists, regarded all erroneous cognition as arising in mental misconstrual of sensation. 23 To consider the stock Indian example of erroneous cognition, seeing a mirage and taking it to be water rather than distant heat waves rising from the earth, Diimaga would say that it is not at the level of sensation that error occurs, for we really do sense something, and we are not mistaking a sensum for something that is not a sensum. Rather it is at the level of making a judgement that the error occurs. The error consists in the mind's imposing upon a visual sense datum a concept that later turns out to have been the wrong one. 24

4.2.1.1 Awareness's awareness of itself It was said above that all acts of awareness are sense data, since acts of awareness constitute the field of operation of a sense faculty, namely, the mind (manas). But actually the matter is not quite this straightforward in Diimaga's system, for the mind has a very different status from that of the other sense faculties. Whereas the five physical sense faculties are regarded as separate entities that exist apart from the objects of which they are aware, the mind turns out in Diimaga's view to be merely an aspect of the acts of awareness that putatively constitute the mind's field of operation. Diimaga argues that each act of awareness is aware of itself, which amounts to saying that the instrument of the act of awareness, the act of awareness itself and the awareness of that awareness are in fact a single entity. Closely related to this doctrine is his doctrine that awareness and the means of acquiring that awareness are also a single entity. These two theses form the subject matter of five verses and their commentary in the first chapter of the PramiifJasamuccaya, 1:8-12. An outline of his argument in those verse will shed some light on Diimaga's views on sensation. First let us begin with an analys~s of a single datum, namely, the fact that a cognition has occurred with a given content. Let us symbolize this datum "K(c)". When we think about this datum K(c), says Dinnaga, we are inclined totry to analyse it into three factors:

1. an object that has been cognized, that is, the content, of the act of awareness, the c of K(c) 2. awareness itself, the K of K(c) 3. an actIVIty, performed by consciousness, of grasping or apprehending the object that becomes its content.

Chapter Four

141

When we analyse our single datum K(c) in this way, we naturally regard the activity of apprehending as an instrumental cause, which we call the pram[lIJa in the sense of instrumental cause (karat;a) of acquiring knowledge; and we regard K(c), the awareness of the object, as pramti1}.a-phaia, that is, an effect of that instrumental cause. We call this effect cognition (jfttina). What prompts us to make this kind of analysis is the fact that when we look back at an act of awareness, we can recall two things, namely the object c that was cognized, and K(c) that is the very fact that we were aware of the object. We may symbolize the recollection of the object itself as "R(c)" and the recollection of the act of awareness "R[K(c)]". Now given these two recollections, R(c) and R[K(c)], it is natural to assume that each of them is the recollection of its own distinct act of awareness, since a recollection is the remembering of an earlier act of awareness that may be called the original cognition. That is, we are inclined to believe that recollection R(c) is based upon the original cognition K(c) and that R[K(c)] is based upon the original cognition K[K(c)]. But the above assumption of double acts of awareness, K(c) and K[K(c)], does not, argues Diimaga, stand up well under close examination. For if we accept the principle that any given act of awareness requires a second act of awareness to be aware of it, we are led into an infinite number of distinct acts of awareness. K(c) would be followed by K[K(c)], which would be followed by K(K[K(c))), and so on indefinitely. To avoid this infinite regress, Diimaga suggests that it is preferable to say that the act of awareness of an object requires no second act of awareness to be aware of it. Cognition of an pbject and awareness of that cognition is a single act. Every single act of awareness is an awareness both of its object and of itself as awareness. Expressed symbolic~lly, K(c) = K[K(c)] = K(K[K(c)]). Moreover, recollection of an object R(c) and recollection of an awareness R[K(c)] are just two recollections about different aspects of a single entity K(c), cognition-cum-content, which appears not to be further reducible into the component K and c. Now in the above attempt to think about K(c) by analysing it into three factors, one of the factors mentioned was an instrumental cause of awareness whereby K apprehended c. According to Diimaga's view there actually turns out to be no such instrumental cause, at least not in the sense of there being a separately existing thing that serves as an instrument. But it is still not entirely meaningless to talk in terms of an instrumental cause of a cognition insofar as, if K(c) is aware of itself, then we may consider K(c) to be its own instrumental cause. Therefore, says Diimaga, the instrumental cause of a cognition is the same entity as the resultant cog-

142

DiIiniiga's Theory of Knowledge

nition. In other words, a pramii1)a, in the sense of a means of acquiring knowledge, is identical with the pramii1)a-phala, in the sense of the knowledge resulting from that means. Therefore, not only is every act of awareness, regardless of whether it is a sensation or a judgement, a sensum, but insofar as it is aware of itself it is also an aCt of sensation. This conclusion reached by Diimiiga, that the terms ''pramii1)a'' and "pramii1)aphala" refer to two aspects of the same entity and that a cognition cognizes itself, drew a considerable amount of criticism from Uddyotakara and later members of the Naiyiiyikas. In the Buddhist camp, however, these conclusions became a matter of orthodoxy, probably because they were quite congruent both with the fundamental claim of Buddhism that there is no experiencing self as a reality separate from the process of experience itself, and with the phenomenalistic trend of the Yogiiciira or Vijfiiinaviida school of Buddhism with whose system of abhidharma the later Buddhist epistemologists tended to be sympathetic. 4.2.2 Inference in the Pramii.1Jasamuccaya With the above topics treated in the opening part of the fIrst chapter of the Pramii1)asamuccaya as background, we can now tum to a discussion of some of the topics in the second chapter of that work. In broad outline at least, I will discuss the topics in the same order as DiIiniiga discusses them. As has already been indicated, the principal distinction between sensation and inference is that the former is a process of being aware of objects present to the senses, while the latter is a process of cognizing objects not present to the senses. The peculiar attributes (sva/a/qa1)a) of things can be cognized through sensation, while inference gives us no cognition of peculiar attributes but only of general attributes. To use the' standard example, when one infers that there is fIre upon observing a body of smoke, the resultant cognition can only be the general knowledge that there is some fIre in a certain place, but it can never be knowledge of which fIre it is or of what sort of fIre it is. And conversely, sensation gives us no general information. Sensation gives us only the most simple cognition of exactly the thing at hand, but it gives no further information· as to what this thing has in common with other things or of which class this sensed object is a member or what this sensed thing is called. Incidentally,on the matter of whether general attributes, universals and so forth could be sensed directly, the Buddhist philosophers in general differed sharply with other schools of Indian philosophy, many of which held that universals are directly sensed rather than superimposed by the mind upon sense data. 25 Now this led to a further distinction between

Chapter Four

143

sensation and inference, which is that sensation is quite private in that a sensation cannot be shared by communicating it to another sentient being's mind. Diimaga writes "No knowledge at all of a possessor of properties that has many characteristics is derived from a sense faculty. The field of operation of a sense faculty is a characteristic that is experienceable by itself and inexpressible. ,,26 An inferential cognition, on the other hand, can be communicated, for it is possible to tell some other person in a general way that which one knows in a general way, and the other person will be able, provided she knows the linguistic conventions one is using, to understand approximately what one is talking about. The structure of the thought process and the nature of the inferentially derived cognition are essentially the same whether communicated to other people or not. It is on these grounds that Dinnaga treats verbal communication as a special case of inference rather than, as was generally the case in other schools of Indian philosophy, as a means of acquiring knowledge distinct from inference. But since communicating our cognitions to others, especially when these others would prefer not to believe what we are communicating to them, requires special techniques and is governed by certain conventions, Dinnaga feels this aspect of inference deserves chapters in his book separate from the chapter that deals with the basic structure of inferential cognition. Therefore chapter two of the Pramaf;asamuccaya deals with these basic structures, chapter five shows that those same basic structures are found in the use of verbal symbols, and chapters three, four and six deal primarily with the conventions of debate and show how the basic structures of inference underlie those conventions. 4.2.2.1 The subject matter of inference If inference is regarded as a means of acquiring knowledge, the question naturally arises what the content of that new knowledge is. In dealing with this question, Dmnaga considers two answers that one might put forth, rejects them as inadequate, and offers a third answer of his own. The first position he considers is that from the observation of one property we gain knowledge simply of a second property. Most likely what Dinnaga had in mind in discussing this position was that style of inference that deals primarily with causes and their effects, whereby a cause can be inferred from the observation of its effect. The period of Indian philosophy before Diimaga's time had been one of considerable preoccupation with questi,ons of various kinds of causation. The question of causation made up a substantial part of Buddhist abhidharma literature and of the literature of the systems of natural philosophy such as the Sarhkhya school and the V aise~ika school. All these schools of thought tried to determine the most

144

Diimaga's Theory of Knowledge

fundamental kinds of cause, for all shared the view that happiness could be attained only by understanding the cause of unhappiness and then eliminating that cause so as to eliminate the effect. The whole notion of the possible relationship between cause and effect was subjected to close scrutiny by Nagarjuna and the early Madhyamaka school, with the result that causality came to be regarded as a conventional truth but one that would not stand up to the rigours of logical scrutiny. Perhaps because of the tendency of early Mahayana thinkers to be skeptical of causality, the early Buddhist epistemologists tended not to discuss causal relationships as relevant to the process ofinference. But the Vaise~ikas had dealt fully with questions of inference based on causal relations. They dealt, for example, with predictions of future effects from present causes and knowledge of past causes from present effects. 27 In inferences of this type the new knowledge that one is supposed to attain on seeing an effect is knowledge of the cause. But DiIinaga rejects this view on the grounds that there is in fact nothing new to be learned in this case. When we infer fire from observing smoke, he points out, the only knowledge we acquire is a very general knowledge of fire. But we already had a general knowledge of fire before we made the observation of smoke, so we learn nothing new by observing a body of smoke. If, on the other hand, we did not already have a general knowledge of fire, it can only be because we have never before experienced it, and if that is the case, then the observation of smoke cannot generate any awareness whatsoever of fire. Therefore it is not the case that from the observation of one property, such as the property of being smoke, we gain knowledge simply of another property, such as the property of being fire. The second position that Diimaga considers is that the object of inference is the relation of the inferred property to its locus or bearer. According to this view, we already know smoke in general and fire in general and the relation between them, but we learn of the relation between fire and the locus of smoke. Thus this relation is the object of inference. The general idea of this view is similar to DiIinaga's view, but he rejects this formulation on the grounds that when we make an inference our knowledge of the relation between smoke in general and fire in general has the form "Every bearer of smoke is a bearer of fire." Given this universal proposition and the proposition derived from an observation "This is a locus of smoke," we can derive "This is a locus offITe." Thus in Diimaga's view it is just the locus of the inferred property that is the subject matter of inference. The relation between inferred property and its bearer cannot be the subject matter of inference, because it is not a bearer of fITe or smoke; rather, fITe and smoke are regarded as the loci of the relation between them.

Chapter Four

145

4.2.2.2 Three characteristics of legitimate evidence The next question to be dealt with is one that arises naturally from all that has been said so far about inference: under what conditions can the awareness of a second property in a given bearer be said to follow legitimately from the observation of the first property in that bearer? We have already dealt with this topic at some length in the discussion of Dinnaga's Hetucakranin;aya, which was a summary of the tradition of formal debate (vada) as Dinnaga had inherited it from his predecessors. According to that tradition, a piece of evidence offered in a debate as a reason for some conclusion could be considered legitimate only if it possessed three characteristics. Dinnaga adapted this test for legitimate evidence in debate, where one is trying to convince others, to the case of epistemology, where one is trying to determine for himself the correctness of a tentative judgement concerning the location of a "hidden" property in a given property-bearer. Let us first examine the classically formulated three characteristics of legitimate evidence in debate. In this discussion I shall use the term "legitimate" to qualify evidence that points only to the conclusion stated in debate and not to that conclusion's negation. The conclusion stated in a debate has the form "A certain property occurs in a given locus." Of course the statement actually used need not have precisely that same syntax, but it should be a statement that expresses the same state of affairs as a sentence of that form. For example, all the following sentences express the same state of affairs, and expressions of these forms can be used interchangeably to express the belief that a certain property occurs in a given locus: 1. The fact of being orange occurs at the nose of the cat.

2. 3. 4. 5.

The property "that it is orange" is in the locus of the cat's nose. Orangeness is in the cat's nose. The cat's nose has an orange colour. The cat's nose is orange.

The last of these examples is the most idiomatic in natural speech, while the others are somewhat artificial ways of drawing special attention to the location of a property in a locus. Since these relatively artificial expressions may have a greater capacity for precision and for removing ambiguity in some contexts, it is important to have the flexibility to use whichever syntactic form is most suitable for communicating an idea in dialogue. Indian logicians, like their Greek counterparts, tended to prefer the relatively artificial "There is orangeness in the cat's nose" to the

146

Dirinaga's Theory of Knowledge

relatively idiomatic "The cat's nose is orange." It is quite possible that this preference stems from considerations of clarity. In his study of Aristotle's formal logic, Jan Lukasiewicz notes that Aristotle preferred expressions of the form "Science belongs to all medicine" to the more natural "All medicine is science." His commentator Alexander explained that this artificiality was deliberate. Lukasiewicz summarizes Alexander's comments: According to this explanation, in fonnulae with the verb 'to be predicated of something' and, we may add, with the verb 'to belong to something', the subject and the predicate are better distinguishable (gnorimoteroi) than, we may add again, in formulae with the verb 'to be'. In fact, in formulae with 'to be' the subject as well as the predicate is used in the nominative; in formulae preferred by Aristotle only the predicate is in the nominative, and the subject is either in the genitive or in the dative, and therefore can be more easily distinguished from the subject. 28

Exactly the same observation can be made about sentences in Sanskrit, in which two words in the nominative case occurring in a sentence can easily give rise to an ambiguity as to which is the subject and which the predicate, an ambiguity that arises since the order of the words for subject and predicate is not fixed; this ambiguity is removed in a Sanskrit sentence in which the predicate is in the nominative and the subject in the locative or genitive case. Moreover, since the subject is normally expressed in the genitive or locative case in the context of discussions of logic, it is convenient to refer to the subject as a property possessor (dharmin) or a property locus, where the word "locus" is a translation of the technical term "adhikarafJa," which in Sanskrit grammar is used to name what is conveyed by the seventh case and in philosophical discussions used in the sense of substratum or subject in which a property occurs. In the context of a formal debate the property whose occurrence in a given locus is being argued may be called the arguable property (sadhya), and the given locus in which the arguable property putatively occurs may be called the inferable object (anumeya) or the subject (pak$a) of the inference. The evidence (hetu), also frequently called the inferential sign (linga), is to be understood as that property from which knowledge of the occurrence of the arguable property in the subject is derived. This evidence can be considered legitimate only if all of the three following characteristics occur in the property (compare p. 122): 1. The inferential sign must be a property of the subject of the inference. 2. The inferential sign must be known to occur in loci, other than the subject of inference, in which the arguable property occurs.

Chapter Four

147

3. The inferential sign must not be known to occur in other loci in which the arguable property is absent.

In the context of debate, the first characteristic rules out the introduction of irrelevant evidence that has no connection with the subject of the debate. The second characteristic rules out two kinds of poor evidence. First it rules out evidence that points only to the negation of the stated conclusion. And second it rules out those properties that occur in no locus other than the subject. And the third characteristic rules out ambiguous evidence, that is, evidence that could point either to the arguable property or to its absence. Those are the three characteristics of legitimate evidence as Dinnaga inherited them. 29 The traditional statement of the second and third characteristics leaves room for some doubt, for the statements are not quantified. It is not, for example, clear whether we are to take the statement of the second characteristic as a universal or as a particular statement. Should it be interpreted as saying "The inferential sign must be known to occur in all loci, other than the subject of inference, in which the arguable property occurs"? In fact, Diimaga realized that that interpretation would lead to an unnecessarily strong statement, one that would eliminate inferences such as "Sound is transitory, because it is manmade." In this inference, the inferential sign is the property of being manmade. Obviously this property does not belong to everything that has the arguable property, transitoriness, for many transitory things are not manmade. Or, to take an example of a syllogistic form from the European tradition, the overly strong statement of the second characteristic would eliminate this inference: 1. All human beings are mortal. 2. All citizens of Tegucigalpa are human beings. 3. Therefore, all citizens of Tegucigalpa are mortal.

The inferential sign, the property of being a human being, does not belong to everything that has the arguable property, mortality. In order to ensure that the statement of the second characteristic of legitimate evidence not be taken in this overly strong sense, Dinnaga refined the statement of the second characteristic as follows: The inferential sign must be known to occur in at least one locus, other than the subject of inference, in which the arguable property occurs.

The statement of the third characteristic, on the other hand, must be interpreted as a universal proposition in order not to admit a large number of arguments that are clearly faulty. For if the third characteristic were "There must be at least one other locus in which the arguable property is

148

Diimaga's Theory of Knowledge

absent in which the inferential sign is known not to. occur," this 'argument would be acceptable: "Sound is eternal, because it is incorporeal." For the arguable property, being eternal, is known not to occur in pottery in which the inferential sign, the property of being incorporeal, is absent. Similarly, this argument would be acceptable: 1. Some non-birds do not live in Canada. 2. All crows are birds. 3. Therefore, all crows live in Canada.

The arguable property, the fact of living in Canada, is absent in at least one thing that lacks the inferential sign, the property of being a bird. In order not to allow obviously faulty arguments such as these to pass as reasonable, Diilnaga reformulated the statement of the third characteristic of proper evidence to read: The inferential sign must not be known to occur in any other loci in which the arguable property is absent.

This refinement in the statement of the three characteristics of legitimate evidence was achieved by the introduction of the restrictive particle "eva" into the Sanskrit sentences stating the last two characteristics.3D Since this innovation is rather important, it is worth considering its development in some detail here before proceeding any further. The passage in which the use of the particle "eva," translated into Tibetan as "kho na," occurs is in Diilnaga's own commentary to Pramiif)asamuccaya 2:5cd. That discussion begins by referring to a phrase in Pramiif)asamuccaya 2: 1 that reads "Inference for oneself consists in discerning an object through a sign that has three characteristics." The phrase "through a sign that has three characteristics" must be explained. [The sign is] present in the inferable object and what is similar 10 it and absent in their absence. The inferable object is a property-bearer qualified by a property. After observing [the sign] there, either through sensation or through inference, one confirms that it is also present in a general way, either wholly or partially, in what is of the same class. Why is that? Since the restriction is such that [the sign] occur in only what is similar, there is no restriction that it only occur. But in that case it could be argued that nothing is accomplished by saying "it is absent in their absence." This statement is made in order to emphasize that [the sign], being absent in the absence [of subjectlike objects] is not present in what is other than or incompatible with the inferable object.

Chapter Four

149

In this seemingly simple passage Diimaga set in motion, probably unwittingly, a great.deal of discussion over the best way to formulate the nature of the three characteristics of legitimate evidence. One philosopher who responded to DiIlnaga's introduction of the simple word "only" (eva; kho na) into the formula that had come down from the earlier tradition was Uddyotakara. This sharp-minded critic took great pains to point out some of the disasters that Diimaga is courting by introducing the restrictive particle into his discussion of the three characteristics of legitimate evidence. Let us first look at how Diimaga uses the particle, then turn to Uddyotakara's criticisms. In the passage under discussion Diimaga has stated the second characteristic as follows: "The sign is present in what is similar [to the inferable object]. ,,31 As it stands, this is an indefinite proposition. It can, in principle at least, be restricted in one of three ways. First, the subject "sign" can be restricted to the predicate "present in what is similar." This allows that the extension of the predicate may be wider than the subject, and it disallows that the negation of the predicate be true of the subject. What this restriction says, in other words, is that it is not the case that there exists any locus L similar to the subject of the inference such that the inferential sign is not present in L; thus, the inferential sign is present in all loci that are similar to the subject of inference. This universal proposition would be written in Sanskrit: "liilgasya tattulye sadbhaval). eva. (The inferential sign is only present [that is, is never absent] in what is similar.)" Diilnaga explicitly says that he does not intend the proposition to be restricted in this manner, for it would, as Uddyotakara points out, eliminate inferences of the form "Sound is transitory, because it is manmade." Second, the full predicate "present in .what is similar" can be restricted to the subject "sign." This allows that the subject may be wider than the predicate. What this restriction says is that it is not the case that there exists any property other than the inferential sign that occurs in both the subject of the inference and what is similar to the subject in virtue of possession of a second inferable property. This universal proposition would be written in Sanskrit "liilgasya eva tattulye sadbhavai).. (Only the inferential sign [and no other property] is present in what is similar." This restriction is hardly even worth the effort it takes to point out that it is selfcontradictory, since it is obvious that if the subject of inference has only one property, namely, the inferential sign, it cannot have a second property to be inferred through the inferential sign. Diilnaga apparently saw no need to mention this second possible interpretation of the restriction.

150

Dirinaga's Theory of Knowledge

Third, the subject and predicate "the sign is present" taken together can be restricted to the qualifying phrase "in what is similar." This allows that the extension of "what is similar" may be wider than the extension of "the sign's presence," and it disallows that the sign's presence be found in any locus of the set of loci complementary to the set of loci to which "is similar [in virtue of being a locus of the inferable property] to the subject of inference" is truly predicable. What this restriction says, in other words, is that it is not the case that the inferential sign is present in some locus L such that L is not similar to the subject of inference with respect to possession of the inferable property. But this restriction does not imply that the inferential sign is present in every L such that L is similar to the subject of inference. This is the restriction that Diimaga explicitly prescribes be read into his formulation of the second characteristic of proper evidence. This, however, gives rise to the question that Diimaga next addresses: "But in that case it could be argued that nothing is accomplished by saying 'it is absent in their absence'." The second characteristic of proper evidence apparently makes the third characteristic redundant. About this more will be said later, but let us turn now to Uddyotakara's criticisms. The main theme of Uddyotakara's criticisms is that Dit'maga has been unjustifiably careless in his introduction of the restrictive word "only" into his interpretation of the three characteristics. The gist of Uddyotakara's attack is as follows: 1. Diimaga has said nothing about whether or not the inferential sign must occur throughout the subject of inference. If characteristics two and three are to be quantified rather than left indefinite, then why should characteristic one not be quantified? Diimaga has taken care to spell out that a property can be used as evidence so long as it does not occur in any other loci in which the arguable property is absent and so long as it does occur in at least one locus, other than the subject of inference, in which the arguable property occurs. But no mention has been made of whether the inferential sign must reside in all the subject of inference or whether it is sufficient that it reside in only apart.

2. Diimaga has specified that the inferential sign's presence be restricted to what is similar to the subject of inference. But this surely eliminates the subject of inference itself from inclusion in the set of loci in which the inferential sign is present, for the subject of inference is not among those loci that are similar to the subject of

Chapter Four

151

inference. Thus the second characteristic of legitimate evidence, as Diimaga interprets it, apparently contradicts the first. 3. It is unnecessary to state both characteristic two and characteristic three as Diimaga has reformulated them. For the formulation of the second says that the inferential sign can occur nowhere but in loci in which the arguable property occurs, and the formulation of the third characteristic says, redundantly, that the inferential sign cannot occur in loci in which the arguable property is absent. Now it is clear from the context of Diimaga's discussion what he intended to accomplish by introducing the restrictive word "only." He intended to justify his claim that an inferential sign can still be legitimate even if it resides in only some members of the set of loci that are similar to the subject of inference, bui that it must be absent from all dissimilar loci. The importance. of Uddyotakara's criticisms lies in his successfully pointing out that what Diimaga actually said is not entirely consistent with what he intended to say. The intention of stating the third characteristic, according to Diimaga's interpretation, is to rule out as proper evidence any property that occurs in a locus in which the arguable property is absent. It is noteworthy that, in interpreting the third characteristic in this way, DiiJ.naga has not adequately answered the criticism that he himself has anticipated here, namely, the question of what this third characteristic stipulates that is not already provided for by the second characteristic. He sets out the problem clearly enough, but his answer to the question lacks the same degree of clarity. The problem bears looking at still more carefully. It appears at first that in the prose commentary to PramiifJasamuccaya 2:5cd, characteristics two and three of legitimate evidence are to be reformulated in a different way than they were formulated in the HetucakranirfJaya discussion. For according to PramiifJasamuccaya 2:5cd we are to understand the statement of the association (anvaya) relation as

being a statement that says two things: 1. The inferential sign does occur "generally" with possessors of the arguable property, as opposed to being restricted to occurring with S-possessors that belong to the class P that is the subject of the inference; in other words, -PHS>O. 2. The inferential sign occurs only with possessors of the arguable property, which means that it does not occur anywhere without the arguable property; in other words, -PH-S=O.

152

Diimaga's Theory of Knowledge

If the association relation is reformulated in this way, it alone says all that used to be said by the statement of the association relation and the statement of the dissociation relation together in the HetucakraninJaya formulation. If this is the correct interpretation of the reformulated second characteristic of proper evidence, how is the reformulated third characteristic to be interpreted? Two possibilities suggest themselves. First, it could be that the third characteristic, which describes the dissociation (vyatireka) relation, is indeed redundant and does no more than to repeat part of what is intended by the reformulated second characteristic, which states the association (anvaya) relation. Second, it could be that the dissociation clause adds one new stipulation to what was stipulated by the original formulation. Whereas the original had said "the inferential sign must not be known to occur in any other loci in which the arguable property is absent," the reformulated third characteristic might be construed as saying "the inferential sign must not be known to occur in any loci in which the arguable property is absent." Indeed, adopting this second interpretation of the third characteristic, whereby it specifies H-S=O, has as a consequence that the second characteristic becomes altogether superfluous, for it would be possible to derive the conclusion P-S=O from P-H=O and H-S=O alone, without any reference to the association clause that says -PH-S=O. What stands in the way of our accepting this interpretation is that Diimaga clearly does not ever regard the association relation as superfluous. On the contrary, he insists that it be in place in order to disqualify one form of argument that would be considered good in the absence of the second characteristic of legitimate evidence. Diimaga insists at Pramiil]asamuccaya 2:7 that arguments of the form "Sound is transitory, because it is audible" are inconclusive. It is DiiJ.naga's claim that the inferential sign, the property of being audible, does not compel us to accept that every sound has the arguable property, whatever that arguable property might be. The key question is: why are arguments of this form not conclusive? Now if the association and dissociation relations are understood as being merely contrapositives of one another, then what we should expect to find DiiJ.naga saying to account for the inconclusiveness of such arguments is something equivalent to the observation that although the argument is formally valid, one of the premises is false and therefore the argument is unsound. Pursuing this possibility, let us see which premise could be false. The premise "no sound is inaudible" is true by definition, for being audible is precisely what is meant to say of something that it is a sound. Therefore, the only premise that could possibly be false is "No audible thing is non-transitory." How could this be shown to be false? It can be shown to be false if there do exist any sounds at all that begin and end. Producing an instance of a

Chapter Four

153

transitory sound should be easy enough, and so we should expect that if Diimaga ruled out arguments of this form on the grounds that a premiss of a formally valid argument was false, he would have pointed out that the reformulated dissociation relation, symbolized H-S=O, does not hold in the argument "Sound is transitory, because it is audible." But this is not in fact what Diimaga does. Rather, he says the argument is not good because it does fulfill the dissociation relation but does not fulfill the association relation. And so, we can only conclude, the association relation as described by the second characteristic of proper evidence is not superfluous after all, and the second and third characteristics of proper evidence are not in a position of contraposition to one another. Since DiIinaga's statements on this issue are in a state of conflict with one another, we are not in a position to state definitively what the proper interpretation of DiIinaga' s formulation of the three characteristics oflegitimate evidence is. His lack of clarity on this matter gave rise, as was mentioned above, to several centuries of bickering among his various interpreters. What we have considered so far are the characteristics for proper evidence in the context of argument, where one is systematically communicating to another person the conclusions one has reached from certain observations. The exact interpretation of the characteristics may be in doubt, but we can at least know that in DiIinaga's view the applicability of these three characteristics to privately reached judgements involves only substituting in the above formula the word "infer" for "argue." This will give us the formula for judgements derived from the observation of evidence. A more general formula for the test of the reliability of any judgement of the form "a certain hidden property is in a given locus" might in DiIinaga's system of epistemology appear as follows: A judgement is reliable only if the following three characteristics are in place. 1. The inferential sign must be a property of the subject of the inference. That is, there exists in the subject of inference a property, which is different from the inferable property and which is furthermore evident to the person drawing the inference; this second property may serve as an inferential sign in case it has two further characteristics. 2. The inferential sign must be known to occur in at least one locus, other than the subject of inference, in which the inferable property occurs. 3. The inferential sign must not be known to occur in any other loci in which the inferable property is absent.

154

Diimaga's Theory of Knowledge

With these few modifications, the transition from the pre-Diimagan study of evidence fit to be used in debate to the Diimagan study of an inferential sign fit to be used in drawing a private inference is complete. 4.2.2.3 On errancy and pervasion From what we have seen so far concerning Dirinaga's Hetucakranin;aya and the discussion of inference in the PramGfJasamuccaya, it should be apparent that underlying Dirinaga's system of logic there was an awareness of logical principles similar to those worked out by traditional European logicians as the logic of propositions. But it should be borne in mind that neither Diilnaga nor his successors in the Buddhist tradition ever worked out an explicit statement of these principles of formal logic, nor did they develop a vocabulary of technical terms corresponding to such terms in European logic as universal proposition, particular proposition, singular proposition, and so forth. Rather, they worked out a set of technical terms that were suited to the task of describing the various kinds of relations that might obtain between one property and another or between a property and its bearer. We have already had a preliminary look at these relations in the context of the discussion of the HetucakranirfJaya, but let us turn now to seeing how they are handled in the PramGfJasamuccaya. The most important principles in Diilnaga's system of relations are to be found in verses 12-25 of the second chapter of his PramGfJasamuccaya. Here he begins with his discussion of a property used as a sign of the occurrence of a second property in the same locus. Now as we have already seen above (p. 138), in Dirinaga's system a "property locus" or "property-bearer" (dharmin) is a conceptual construct rather than a sensum. It is a fiction useful for analysis and understanding, but it does not correspond to any realities in the world as it stands independent of our awareness of it. Once it is allowed that the principal constraints on our use of conceptual fictions is that those fictions serve a purpose in our understanding and that we take care to bear in mind that they are fictions rather than representations of realities, we have a great deal of conceptual freedom and flexibility in how we go about accounting for our experiences. Thus that which is a property of a certain locus can in turn be regarded as a locus of another property or properties, for it is really our thinking alone that makes experientia into properties or property-loci or relations or relata. For example, smoke can be regarded as a property of a smoky locus, but it can also be regarded as the locus of a number of properties of its own. Some of these properties of smoke can be found to occur in many loci, and some can be found in only a few, Still other properties can be found in only one specific body of smoke. Now given

Chapter Four

155

this abundance of properties that can be present in a specific locus, the question naturally arises, when that locus is taken as a property of yet some other locus and used as evidence for some further property in that locus, which of the many properties on hand can serve as signs for properties outside themselves? In dealing with this question Diimaga establishes fIrst of all that properties that are unique to a locus can by themselves lead to no knowledge of properties beyond themselves. They are sensa that cannot be relied upon as signs. They must be assisted by recollections of past experiences, which provide associations pertinent to the particulars that are being sensed. That is to say, we identify the particular sensa or classify them according to past experiences. This process of identification is itself rather complex, for it involves a series of judgements that consist in attributing increasingly narrow classes to the particulars at hand. If we were to recreate an artifIcial approximation of the process that we go through in identifying a sensum as a cloud of smoke, we arrive at something like this: "Visible datum. Insofar as it is not absent, it is a presence. Being a presence, insofar as it is not an action or a quality, it is a substance. Being a substance, insofar as it has qualities that belong to smoke but not to other substances, it is smoke." Now in this process of identifIcation a number of properties have been associated with the particular sense datum, and these properties are expressible by such words as "present," "substance," and "smoke." But which of these properties is significant when their locus is itself regarded as a property serving as an inferential sign? In answer to the above question, Diimaga says that of the properties of the inferential sign, only those that are not erratic with respect to the inferable property are capable of playing a role in the inference of that inferable property. Not being erratic means not occurring in loci other than loci of the inferable property. As for properties of the inferable property, only those that occur in every known locus of the inferable property can be indicated by the inferential sign.

The Sanskrit word translated as "erratic" is "vyabhicarin," which comes from the root vi+abhi+CAR, which means to move (CAR) in the direction opposite to (vi) towards (abhi) here, that is, to move away. It is commonly used as a sexual term in the sense of straying from one's rightful partner, being sexually unfaithful or promiscuous, and even more generally being unreliable. The English word "erratic" also means wandering or straying and can, like "vyabhicarin," carry connotations of erroneousness. 32 The condition of one property X being able to occur in

156

Dinnaga's Theory of Knowledge

loci in which another property Y is absent is expressed in Sanskrit by the abstract noun "vyabhicara" or by the finite verb "vyabhicarati." The entire notion "X can occur in loci other than those in which Y occurs" is expressed by a sentence of the form "Xl y5 vyabhicarati" or "X6 y5 vyabhicaral)," where the superscript numbers refer to the numbers conventionally applied to case endings by the Sanskrit grammarians.33 But if X is not erratic with respect to Y, that is, if X is restricted to Y, this restriction is expressed by "avyabhicara" or "na vyabhicarati." Thus all that we have said so far concerning which properties can be inferred from which evidence can be expressed as follows: 1. If "Xl y5 vyabhicarati" is true, then Y cannot be known for certain from knowledge of X. 2. Y can be known for certain from knowledge of X only if "na Xl y5 vyabhicarati" is true (that is, "Xl y5 vyabhicarati" is false). That is how Dinnaga formulates the principle that from knowledge of any given class, one can derive knowledge of any wider class that includes it but not of that wider class's subclasses. The concept of errancy is related to the concept of restriction in that restriction is the contradictory of errancy. Dinnaga introduces another concept that is related to the concept of restriction, namely, the concept of pervasion (vyapti). A property X pervades a property Y if X occurs in every locus of Y. To see how pervasion relates to restriction, let us imagine a small universe that is made .up of only four loci (a, b, c, d) and four properties (W, X, Y, Z). Let us imagine that the four properties are distributed in the four loci as follows: a

b· c d

W,X, Y, Z W,X, Z W,X,Y W,X

In this universe we can observe the following cases of pervasion:

1. W pervades X 2. W pervades Y 3. W pervades Z

4. X pervades W 5. X pervades Y 6. X pervades Z

And we can also observe the following cases of errancy:

Chapter Four

157

1. W is erratic with respect to Y. 2. W is erratic with respect to Z. 3. X is erratic with respect to Y. 4. X is erratic with respect to Z. 5. Y is erratic with respect to Z. 6. Z is erratic with respect to Y. Observation of the above instances shows that pervasion is a nonsymmetrical relation in the four-locus universe. Wand X are in a relation of reciprocal pervasion, but X and Yare in a relation of non-reciprocal pervasion in that X pervades Y but Y does not pervade X. Similarly, errancy can be either reciprocal or non-reciprocal. In the above universe Y and Z are mutually erratic, but Wand Y are non-reciprocally erratic in that X is erratic with respect to Y but Y is restricted to X. From the above it can be seen that given any two properties (PI' P2), there cannot be between them both a relation of reciprocal pervasion and of reciprocal errancy. But it may be that there is neither a relation of reciprocal pervasion nor of reciprocal errancy. In case there is neither reciprocal pervasion nor reciprocal errancy, there must be a relation of non-reciprocal pervasion. If there is non-reciprocal pervasion, there is also non-reciprocal errancy. For PI pervades P2 iff [if and only if] P2 is restricted to PI' And P2 does not pervade PI iff PI is not restricted to P2' Therefore, (PI pervades P2 and P2 does not pervade PI) iff (P2 is restricted to PI and PI is not restricted to P2). The left side of this bi-conditional describes non-reciprocal pervasion, the right non-reciprocal errancy. What follows from these considerations is that between any two properties there must be exactly one of the following three relations: reciprocal errancy, reciprocal pervasion or non-reciprocal pervasi(~m. Let us now return our attention to the problem of inference, the process wherein observation of one property in a locus leads to knowledge of a second property in that locus. It was pointed out above that Dirinaga . laid down the principle that observation of one property, the inferential sign, can lead to knowledge of a second property, the inferable or arguable property, only if the evidence is restricted to loci of the inferable property. And we have seen in the discussion immediately above that one property is restricted to a second property if and only if it is pervaded by that second property. That which is pervaded is necessarily restricted to the pervader, but the pervader is not necessarily restricted to that which is pervaded. Similarly, legitimate evidence can lead to knowledge of the inferable property; but no certainty res\llts if the roles of the properties be reversed, that is if we try to use what was formerly the inferable property as

158

Diimaga's Theory of Know ledge

evidence for the property that was formerly the evidence. The relation of evidence and inferable property, or between sign and thing signified, is a non-symmetrical relation. In all cases of pervasion (vyapti), absence of the pervading property is restricted to absence of the pervaded property. Put symbolically, if we let the pelvading property be called S and the pervaded property H, that statement amounts to this: -S--H=O, or H-S=O. That is the state'%eni of the dissociation (vyatireka) relation. Now any property can serve as a basis for dividing the universe into two sets of loci: the set in which tvat property is present and the set in which that property is absent. So another way to state the dissociation relation is this: "presence of a pervaded property in a locus precludes that locus being a member of the set of loci in which the pervading property is not present." The set of loci in which any given property is absent is called that property's counterextension,} or in Dinnaga's terminology, the property's other (anya). The notion of preclusion is conveyed by the abstract noun "apoha, " which is a nominalized form of a verb that means to deny or to exclude. So the compound formed by these two elements, "anyapoha," refers to the principle of precluding the complement. This principle, according to Dinnaga, describes the essence of the inferential process, for it is that which is shared by private inference and by communication of one's ideas to others, whether that communication be in the form of a formal debate or informal conversation .. In the context of private inference, "preclusion of the complement" refers to preclusion of the membership of the locus of evidence in the counterextension of the inferable property. In the context of verbal communication, it refers to preclusion of a linguistic sign being used to stand for an item in the complement of the set of items to which the sign is correctly applicable according to the speech conventions of the person using the linguistic sign.

4.3 The skepticism implicit in Dilinaga's epistemology In treating Dmnaga's views on reasoning most scholars have made the assumption that Dinnaga's aim was to set forth a set of guidelines by following which one could add to one's knowledge through the inferential process. The prevailing view has been that in Dinnaga' s system there are two bodies of knowledge, namely, knowledge of that which is immediately present to the senses and knowledge of things that are not present to the senses but that can be known because of their invariable relationships to

Chapter Four

159

things that are present to the senses. If we work on this assumption, we are inclined to see the guidelines that Diimaga discusses as a set of sufficient conditions that if met will result in the growth of our knowledge. Making this assumption, however, leads to some rather odd consequences that can be avoided by not making it. The principal odd consequence that arises from making the assumption just outlined is that Diimaga's canons of knowledge turn out to be much more strict than we might at fIrst expect. And if attempts are made to make his canons less strict, then it seems the result is a set of canons that are far too lax. This "failure" to arrive at what seem to our intuitions to be reasonably strict canons of knowledge could be attributed to the fact that Diimaga was after all a pioneer in the field of epistemology and like most pioneers failed to get everything right the first time round. Indeed, some of the awkwardness and lack of precision that we find in his account of the three characteristics of legitimate evidence suggest that Diimaga was not fully in command of some of the concepts that eventually became commonplace in the very school of thought that he is given the credit for founding. Who would fInd it surprising that even a great thinker would leave room for improvements? But another account of Diimaga's "failure" is possible, which is that he did not fail to do the task we have assumed he was trying to do but instead succeeded in doing some other task altogether. Exploring this possibility may lead us nowhere, but if it does, then we shall at least be the wiser for knowing which path not to take in future research on Diimaga. So what I shall do here is to put forth an argument for the hypothesis that Diimaga was a rational skeptic, look at some of the problems of interpretation this hypothesis could solve, and finally examine some of the implications of that position for Buddhist practice as Diimaga might have seen it. But before doing any of that, let me spell out the central issue that remains unclear in our present understanding of Ditmaga' s work. H.G. Herzberger has succeeded in showing that the system of reasoning that I claim is Diimaga's leads to some surprising consequences. Herzberger (1986) deals at length with a system of logic that he calls HD, for Hayes-Dinnaga. His source for the system HD is an early draft of Hayes (1986) in which I give essentially the same account, with only a few minor modifIcations, of the HetucakraninJaya as the one found in the early parts of this chapter. Herzberger has correctly pointed out that the system HD contains a very serious flaw that discredits it as a system of logic. Let me begin, then, with a consideration of ways in which the Hayes-Dinnaga system of logic appears to be at variance with what we normally expect a good system of logic to be.

160

Diimaga's Theory of Knowledge

A system of logic is good to the extent that the formal arguments that its rules sanction correspond to informal arguments that we know from common sense to be good arguments. 34 Using this as a rough guideline of what we look for in a system of formal logic, Diimaga's system of logic is not very good, because it seems to sanction far too many arguments that our common sense would not recognize as good. Herzberger has offered several arguments that meet the three characteristics of legitimate evidence but nevertheless yield false conclusions. Some of those arguments, cast in the form that Indian forms of argument would cast them, are as follows: 1. Cetaceans are land-dwellers, because they are mammals. 2. Monotn:mes are viviparous, because they are mammals. 3. Odd integers are divisible by two, because they are integers. 35 The zoological order of cetaceans comprises sea-dwelling mammals such as dolphins and whales. If what we knew of cetaceans was just that they are mammals, and if the system of reasoning that we had to rely upon was Diilnaga's, then we should begin by establishing the induction domain, which would comprise the class of non-cetaceans. And in the induction domain we should see that there are things, for example coyotes, that possesses the sign, the property of being mammal, and the inferable property, the property of dwelling on the land. And we should fail to find any non-cetacean that was a mammal and did not live on the land. So we should conclude that cetaceans live on the land. Similarly, we should conclude that an animal belonging to the subclass monotreme (to which the Australian duck-billed platypus belongs), which is subclass of the zoological class Mammalia, gives live birth to her young, because all mammals in the induction domain give live birth; we should be unprepared for the fact that monotremes lay eggs. What is to be learned from all these examples of faulty arguments that nevertheless satisfy the Hayes-Diilnaga system of logic is that if we include in that system some form of inductive assumption, that is, the assumption that whatever relations hold in the inductive domain also hold in the subject class of the inference, the result is a system that is, to use Herzberger's term, inductively wild, for the system would allow not just a few but a very large number of bad arguments. As Herzberger states the matter: The general pattern should be abundantly clear. It depends on regularities which have exceptions. Whenever some term extends over all exceptions to a given regularity, a counterexample to fA [the inductive assumption] can be constructed. These are quite commOl! in biology, in arithmetic, in grammar, and can be found in many other

Chapter Four

161

domains as well. No doubt some cases of· this kind could be found involving pots, sounds and other favourite Buddhist terms. 36

Having pointed out the serious weakness of the inductive assumption attributed by me to Diimaga, Herzberger recommends an "attitude of restraint in actually attributing ... to Diimaga" the lines of reasoning that I have suggested are implicit in Diri.naga's work. "The inductive assumption" says Herzberger, does not stand up well under examination, and this leaves us with the problem of explaining how or why Ditmaga might have been committed to it, if he was indeed so committed ... .In point of fact, [Hayes] does not cite any direct textual evidence for attributing fA to Diimaga. This would not pose a serious problem if fA were a reasonable principle, for then it could be justified as a reasonable interpolation. However, fA is notably tenuous. If there were clear direct textual evidence, that might tend to force the attribution, probably to Diimaga's discredit as a 10gicianY

This cautionary note is well founded, and it must be emphasized that much of what follows is speculative in nature and is motivated by a curiosity in seeing whether or not an interpretation can be made of Diri.naga's views on epistemology that is relatively consistent with universal intuitions about the nature of logic, consistent with what Diri.naga himself explicitly says and consistent with the trends of Buddhism in Diri.naga's position in history. Let it also be emphasized that I see no good reason to remain consistent with such interpreters of Diri.naga as Dharmaklrti and his followers, for their interpretations of Diri.naga's views were as anachronistic as those of any twentieth century scholar and may in fact have been even less well informed. Before indulging in any sort of speculation, however, I should address the problem of whether the fatal flaw in the Hayes-Diri.naga system was put there by Diimaga or by Hayes. The explicit form of the inductive assumption, "Assume that the counterpart of whichever compartment of the class of possessors of the evidence is empty in the induction domain is also empty in the subject domain," cannot be attributed to Diri.naga. The motivation for searching for such an assumption was that, as was pointed out above (p. 122), the three characteristics of a proper inferential sign as set down by Diri.naga are not sufficient to ensure the certainty of the conclusion drawn from an inferential sign that has those three characteristics. And yet it appears to be the case that Diri.naga believed that arguments making use of inferential signs that had the three characteristics were good arguments. In fact, he explicitly says of such arguments in the Hetucakranin;aya that they are correct; the term used in the Tibetan translation of HetucakraniTl;aya is

162

Dinnaga's Theory of Knowledge

"yang dag," which is most probably a translation of the Sanskrit "samyafic," which means right, correct, proper, suitable, fitting,

appropriate, good, true and so forth. An example of an argument that Dinnaga certifies as correct is "Sound is transitory. Because it is created. Consider pottery and ether. Present in the subjectlike class and absent in the unsubjectlike class, the evidence is good." Taken absolutely at face value, what this argument says is "There is something that is both created and transitory, namely, pottery. There is something that is neither created nor transitory, namely, ether. Sound is created. Therefore, sound is transitory." Moreover, it is not supposed to be a fluke that this particular argument happened to work out well. Rather, it is presented as a paradigm of an argument which, if substitutions are made for the terms in it, will generate other correct arguments. But it is not at all clear why one should have confidence in arguments of the form "A is an H-possessor and an Spossessor. B is an H-non-possessor and an S-non-possessor. C is an Hpossessor. Therefore, C is an S-possessor." But I hope to have shown that equally important in this form of argument is what is not said. In particular, it is significant that no example is cited of an object that is both an H-possessor and an S-non-possessor. All that is needed is one small assumption in the form "if there were an example of an object that is both an H-possessor and an S-non-possessor, that example would be given" to arrive at the conclusion "no H-possessor is an S-non-possessor." With this conclusion we have the universal proposition that is needed to enable us to derive a conclusion from the observation that the subject of inference is an H-possessor. But the truth of a universal proposition is not secure; it can only be assumed. Finding exactly the form of the proposition whose truth has to be assumed in order to secure the truth of the universal proposition is a useful exercise, for once that assumption is explicitly stated, it is possible to see how very feeble it is. What I claim to have done is to show the form of the proposition that would have to be true in order to make good the arguments Dinnaga endorses. What Herzberger shows clearly is how feeble that assumption is. For in a world in which rules have exceptions, no inference that proceeds from observing a pattern in part of the universe and assuming that the pattern holds in all other parts of the universe is certain. The extent to which Dinnaga may have been aware of the weakness in his own account of inference is something about which it is impossible to decide. What can be said is that he appears to have been less concerned with the question of securing some measure of certainty in inference than was DharmakIrti, for whom the problem was acutely felt. DharmakIrti felt compelled to appeal to the notion of an "essential relationship (svabhavapratibandha)" between the inferential sign and the property inferable

Chapter Four

163

through it. This essential relationship was supposed to hold between certain kinds of objects, in particular between causes and their effects and between natural genera and species. Thus, whereas in DiIi.naga's system there is a great risk involved in assuming that patterns that have been observed so far in one's experience will continue to be found in future experiences, this risk is supposed to be reduced when that pattern is no mere accident but rather is an instance of some essential connection. That DiIi.naga did not appeal to the notion of essential connection has been known since the very outset of the modem interest in classical Buddhist epistemology. As we saw in chapter one, VidyabhilsalJ.a observed as early as 1905 that "The attempt on the part of the Buddhists to divest the principle of logic from those of theology, metaphysics, etc., was the cause of the foundation of the modem Nyaya, otherwise called Tarkasastra or Logic proper.,,38 VidyabhilsalJ.a's observation is not particularly applicable to DharmakIrti, but it certainly fits DiIi.naga. And we also saw that Frauwallner observed that DiIi.naga, unlike the Vaisesikas before him and Dharmaldrti after him, had made no appeal to the notion of causal relations in his study of inference. The question is: how much importance can we attach to DiIi.naga's silence on the question of causal relations? Was his reason for not mentioning them that he felt they had no real place in the study of inference, or was it that he felt their place in inference was too obvious to mention? Insofar as Dharmaldrti presents himself as a commentator on DiIi.naga's work, he seems to have assumed that the appeal to causal relations and natural genus-species relations was implicit in DiIi.naga's account of inference and that Dharma1drti himself was merely making explicit what was implicit in the texts. That is an assumption that may bear up under examination, but at least it should be tested before it is accepted. There are, I think, some considerations that weigh against our reading into DiIi.naga an appeal to causal and natural class relations as a means of securing inductive assumptions. The most important consideration is that DiIi.naga provides no means of knowing about such relations, His account of sensation provides only for having immediate . knowledge of sense data, and he explicitly states that sensation cannot provide knowledge of the individuals that are usually assumed to exist in the external world as a substratum of the various properties that we directly sense. But causation is surely something that can take place only between individuals, not between sense data. It cannot be the case that the sensation of a reddish patch of colour causes the sensation of a greyish patch of colour, but rather it must be the case that a fIre, which is presumably an external object that caused the s"ensation of the reddish patch of colour, also caused the smoke, which is presumably an external

164

Diilnaga's Theory of Knowledge

object that caused the sensation of the greyish patch of colour. But this causal process is one that takes time, and therefore forming the judgement that the causal process has taken place requires not simply an immediate sensation of sense data but also a recollection of previous states of affairs. Such a judgement is a structuring (kalpana) of just the sort that is excluded by Diilnaga from the purview of sensation. But if knowledge of causal relations cannot be known by means of sensation, the only other possibility for gaining knowledge of them available within Diilnaga's system of epistemology is inference. In Diilnaga's account of inference, however, there is no provision for making judgements concerning anything but the presence or absence of properties with one another. Knowledge of causal relations is an essentially richer kind of knowledge than knowledge of mere presence and absence. 39 Knowledge of presence and absence is knowledge that, but knowledge of causal relations is knowledge why. It is, to look at the matter in one way, considerably more difficult to establish a causal connection between events than it is to establish a statistical correlation between them. It requires nothing more than the keeping of records to become aware of the fact that it rains only if there are clouds in the sky. One can know that simple fact and still be quite mistaken as to what, if anything, the clouds have to do with the falling of rain. An illustration of this can be found in debates between Indian philosophers at the time of Diilnaga. All recognized that vision takes place only if there is light, but none of them had any account that was even close to the correct account of what light has to do with the process of seeing. Another example from that time is that it was well known that living beings had brains in their skulls, but the function of the brain was completely misunderstood. The history of science furnishes us with countless instances of keen observation accompanied by explanations that later turned out to be quite inaccurate of why the things observed were as they were. To look at this matter in another way, causal relations require knowledge not only of what is the case but also of counterfactuals, that is, of what would have been the case if what actually is the case had not been the case. Knowing, for example, that fire is a cause of smoke involves not only realizing that there happens to be some fire in the vicinity every time there is some smoke, but also that there would be not smoke if there were no fire. Not only actual states of affairs but also possible states of affairs must be taken into account. To know something of the causal relation between smoke and fire is to know that the occurrence of smoke in the vicinity would not be possible if there were no combustion. I\nowledge of possibility and necessity is an order of understanding much richer than any

Chapter Four

165

that Diimaga takes into account in his relatively lean presentation of the relations among properties that we find in both HetucakraniTl;aya and PramtllJasamuccaya. The principle of charity in interpretation would favour not making Diimaga accountable for the defense of positions that he did not explicitly hold. Since he did not explicitly appeal to a notion of causal relations, he should not be held responsible for the numerous problems that his system of thought would have encountered had he appealed to them. If Diri.naga did not make any attempt to secure his system of inference against the risks of induction, how are we to understand what it is that he was trying to accomplish in his setting down the characteristics of proper evidence? It seems to me that two possible answers emerge. Since these answers are not incompatible with one another, it is not necessary to make a choice between them, but we could accept that Diri.naga was trying to accomplish two things at once. The first possibility is that he was simply offering an account of what kind of judgement is reasonable, where it is understood that being reasonable does not entail being infallible. It would be reasonable, for example, for a person who had observed only non-mammals living in the ocean and had encountered mammals living on the land to infer that whales are not mammals. It would be a matter of drawing a conclusion that turned out to be incorrect, but at least one's error would have been a reasonable one. In a court of law it might be reasonable in the light of the available evidence to return a verdict of not guilty, only to discover later on that the accused person committed the crime after all. That the person subsequently turns out to be guilty does not alter the fact that the verdict of not guilty was a reasonable one given the knowledge available to the jury at the time of its decision. What would be unreasonable in a court of law, or in science, would be to expect that the conclusions of one's best judgement are not liable to revision in the light of new discoveries. To take Diri.naga's favourite topic of controversy as an example, he claims that it is not reasonable to believe that scriptures are infallible sources of knowledge, because they are merely sentences spoken by human beings. When experience shows us that all sentient beings make errors in judgement, there appears not to be any reason to believe that the authors of scripture might be an exception. Nor is there any apparent reason to hold that, when our observation shows us that all speech is produced by living beings, the statements of scripture are not so produced. To believe that something is an exception to a rule is not reasonable unless very good cause can be shown for why and how the exception occurs, or unless we can very directly observe that there is an exception. It would not, for example, be reasonable simply to infer that whales are ocean-dwelling mammals, but when one can see that a whale is

166

Diri.naga's Theory of Know ledge

indeed a mammal, there is no call for inference anyway. But in the dispute over the status of scriptures, there is no possibility of directly seeing that they were never produced, and so it is absurd to believe that they are the one exception to the pattern that sentences are utterances that result from people having ideas that may be incorrect. It may have been Diri.naga's aim simply to establish some guidelines for common-sense rationality. The second possibility for what Diri.naga may have been trying to achieve is to set down the necessary conditions for certainty. On this account, we should understand the three characteristics of legitimate evidence to be saying something like the following. No judgement is certain if it is not based upon evidence that has three characteristics. They are 1. The evidence must be a property of the subject of the inference. 2. The evidence must be known to occur in at least one locus, other than the subject of inference, in which the inferable property occurs. 3. The evidence must be known not to occur in any locus in which the inferable property is absent. There can, of course, never be any piece of evidence that has the third characteristic. But that does not militate against its being a necessary characteristic, just as the fact that no human being has wings does not alter the fact that wings are what a human being would have to have in order to fly without the aid of machimk Diri.naga could, in other words, be stating a canon of certainty that no judgement ever measures up to. His motivation for setting such strict criteria for certainty would then be easily understandable as yet another instance of the general Buddhist caution against holding opinions and making judgements of any kind, without at least an acute appreciation of our lack of justification for holding the opinions we hold . . On a skeptical reading of Di(maga, his message would be to the effect that the only thing that is certain is that we are having the sensations we are having at this very moment. What those sensations betoken is a matter about which we can only speculate. Even the process of identifying what we are experiencing, which we ordinarily take for granted, comes under suspicion. The process of identification itself is normally extremely fast and intuitive, and we are normally only very dimly aware of going into it. We experience a sense datum and think "Tree," or we hear a sound in the night and think "Mice," and the label appears so quickly that we are not usually aware of how very much processing we have done Qf the original datum that lead to our applying the label to it. If challenged how it is that

Chapter Four

167

we know that a noise in the night is a mouse, we are helpless in explaining how it is that we do know. We just know. In much of the daily routine, "just knowing" serves us very well, for it makes it possible for us to function in a world that often moves at a pace faster than we can comfortably think. But there are other times when "just knowing" leads us into some very hazardous terrain. Examples are especially abundant in the areas of religion, where often a great deal of religious "knowledge" is based upon intuitions of seers, mystics and prophets who communicate messages that they "just know" come from some source outside themselves. Part of the function of Buddhist epistemology has always been to enable us to go very slowly through an approximation of the steps of thinking that we ordinarily go though very quickly. It is a process of analysing our intuitions into their basic units, and very frequently the result is to come to the realization that our intuitions were in fact very faulty. Absolutely no intUition is immune from this process of critical analysis. It makes no difference whether the intuition arose in ordinary waking experience, in a dream, in the depth of a hypnotic trance, as part of a meditational experience or as a bolt out of the clear blue sky; the fact remains that the intuition is not worthy of being considered part of our Imowledge until it has passed the rigorous standards of deductive reasoning. Supposing that a skeptical reading is accepted for Diimaga's system of epistemology, the question of the place of logic within Buddhism becomes rather easy to answer. Logic should perhaps not be seen, as was so often done by DharmakIrti and the later Indian traditions and many of the Tibetan traditions, merely as a means of establishing the truth of Buddhist teachings. On the contrary, putting logic to the service of polemics and apologetics is in a way to thwart the very purpose for which it was intended, namely, to counter dogmatism and prejudice. As a weapon in the battle against prejudice that rages in every mind that seeks wisdom-in minds of the vast majority of people who do not seek wisdom, prejudice simply takes full control without a contest~-there is nothing as powerful as the kind of reason that lies at the heart of Diimaga's system of logic. For it should be clear that very few of our judgements in ordinary life pass the standards set by the three characteristics of legitimate' evidence. Taken in its strictest interpretation, none of the judgements of any but a fully omniscient being passes. And, since there is no evidence that there exist any fully omniscient beings, the best available working hypothesis is that no on~'s thinking is immune from errors that require revision in the face of newly discovered realities. Therefore, since we cannot place full reliance on the teachings of any teacher or any tradition of teachers, our only hope in the final analysis lies just in our own resources as individuals. That this

168

Diimaga's Theory of Knowledge

is so is made all the more apparent when it is recalled that comm'unication from one mind to another by means of language is very limited and can never do full justice to realities, as we shall see in greater detail in the next chapter. Furthermore, since every event is strictly spealcing unique and can be found similar to other events only if we disregard distinctions, every reality is a fresh discovery. The process of discovery is sensation, which deals exclusively with what is present (sat) and real (sat). The Sanskrit verb "asti," from which the participle "sat" is formed, conveys the senses of being, existing, taking place, happening, being located somewhere and especially being present (that is, being located right here). Therefore "sat" means not only real or existent in general but more specifically being present. It also conveys the sense of being good. To be good, to be real is, in most Sanskrit philosophical works, to be right here as an object of full awareness. The fresh discovery of the presently real is one of the principal forms of Buddhist mindfulness training. But the fresh discovery of what is immediately present can be made only if the invasion of recollections of past experiences, the very recollection of which is itself often faulty and weighted with prejudice, can be kept under control. And one of the means available for keeping the invasion of memory-driven judgements and prejudgements under control is to apply the rigorous standards of logic to them. When the weakness of our judgements is exposed by this application of the canons of reason, our opinions have a tendency to vanish into nothingness. Or, to use the favourite analogy from the Sarhkhya tradition, when the actress that has been performing her role realizes that the audience knows that the world she has created is an insubstantial illusion, she retires in shame behind the curtain, and the illusion comes to an end; in a similar way, when the world of experience that is created by opinion and prejudice can be revealed for what it is by the light of reason, opinion retires and the world of painful experience gives way to the joy of dispassion. Seen in this way, the task of an epistemologist such as Diimaga is simply to provide one more way of doing so called insight (viptyana) meditation, regarded as crucial for the attainment of dispassion and nirviil}a.

-- Notes --

1. Ajdukiewicz 1974:412.

Chapter Four

169

2. A very similar statement appears in Cohen and Nagel 1962:77, where it is said that a syllogistic inference may be interpreted as "a comparison of the relations between each of two terms and a third, in order to discover the relations of the two terms to each other." This similarity is not surprising, for the tasks of the Aristotelian syllogistic and of the Indian inference are essentially the same, although there are differences in the details of how those tasks are accomplished by the two systems ofreasoning.

3. Here I follow the convention of using the term "extension" of a property or a relation to refer to the set of all objects that have that property or satisfy that relation. And I use the term "intension" of a class to refer to that property or relation in virtue of which the members of a class are indeed members of the class. 4. Venn 1894. See especially chapters two, four and five.

5. Each compartment is either occupied or not occupied, so for each compartment there are just two possibilities. There are four compartments, so the total induction domain has 24 = 16 possible combinations of empty or occupied compartments. 6. The later Buddhist tradition used the term "vipak.<;a" in place of "asapakf;a." 7. Cardona 1967:337. 8. Prior 1955: 127. 9. Venn (1881), chapters six and seven, and Cohen and Nagel (1962:41 ff.) discuss the implicit existential import in the traditional universal affirmative proposition and the lack of implicit existential import in the logic that has developed since the time of Boole. The topic of empty terms, or non-referring expressions, was according to Lukasiewicz (1951:4) not taken up for serious consideration by Aristotle in his logical studies, and indeed the assumption that in expressions of the form "Every A is B" the term A applies to a class that has members is one that runs throughout most of classical logic both in India and in Europe. Therefore in this work I shall always use the term "universal affirmative proposition" in its classical pre-Boolean sense. 10. For a full discussion of this point see Hattori 1968:97-106. 11. Frauwallner 1958 and 1959:96. 12. The Tibetan text for this passage is found in Hattori (1968:177): "mngon sum dang ni rjes su dpag tshad ma dag ni gnyis kho na stel gang gi phyir mtshan nyid gnyis gihal byal

Irang dang spyi'i mtshan nyid dag las gzhan pa'i gzhal bar bya ba med dol Irang gi mtshan nyid kyi yul can ni mngon sum yin lal spyi'i mtshan nyid kyi yul can ni rjes su dpag pa'o zhes shes par bya'ol" The original Sanskrit for these passages has been quoted by Prajfiakaragupta and is given by Hattori (1968:76 and 79, notes 1.11, 1.13 and 1.14) as "pratyak$am anumanam ca pramafJe ... lak$a1Ja-dvayam prameyam. na hi sva-samanyalak~aJ:Iabhyam anyat prameyam asti. sva-lak~aJ:Ia-vi~ayaIi1 hi pratyak~arh samanyalak~aJ:Ia-vi~ayam anumanam iti pratipiidayi~yamaJ:!." The words in italics in these quotations represent passages in verse in the original text.

13. Hattori 1968:177-179. "mngon sum rtog pa dang bral ba shes pa gang la rtog pa med pa de ni mngon sum mol Irtog pa zhes bya ba 'di ji Ita bu zhig ce nal ming dang rigs sogs bsres pa' 01 I' dod rgyal ba'j sgra mams la ming gis khyad par du byas nas don brjod par

170

Notes

byed del Ihas byin zhes bya ba dangl rigs kyi sgra mams la rigs kyis stel ba lang zhes bya ba dangl yon tan gyi sgra mams la yon tan gyis tel dkar po zhes bya ba dangl bya ba'i sgra mams la bya ba'i sgo nas tel 'tshed par byed pa dang I rdzas kyi sgra mams la rdzas kyi sgo nas tel dbyug pa can rva can zhes bya ba Ita bu'ol I'di la kha cig na re 'breI bas khyad par du byas pas yin no zhes zer rol Igzhan dag ni don gyis stong pa'i sgra 'ba' zhig gis don mams khyad par du byas shing brjod do zhes 'dod dol Igang la nag pa de dag med pa de mngon sum mol" KamalasTla quotes the original Sanskrit for much of this passage, which Hattori (1968:83 and 85) cites: "pratyak$am kalpatJlipogham. atha ka kalpana. nama-jaty-iidi-yojanii. yadrccha-sabdesu hi namna visisto 'rtha ucyate clittheti, jati-sabde~u jatya gaur iti, gUl)a-sabde~u gUl)ena sukla iti, kriya-sabde~u kriyaya pac aka iti, dravya-sabde~u dravyel)a dal)QT visal)Iti....anye tv artha-sunyaiJ.! sabdair eva viSi~to 'rtha ucyate." 14. Broad 1923:89. 15. Broad 1923:92-93. 16. Hospers 1953:536. 17. This topic is taken up for discussion by Diimaga in PramafJasamuccaya 1:11-12. See Hattori 1968:29-3l. This topic will also be examined in more detail in 4.2.1.1 below. 18. PramafJQsamuccaya 1:6cd. Hattori 1968: "rnal 'byor mams kyi bla mas bstanj ma ,dres pa yi don tsam mthongl Imal 'byor ba mams kyis kyang lung las mam par rtog pa dang rna 'dres pa'i don tsam mthong ba ni mngon sum mol" The original Sanskrit for this passage is quoted in Hattori (1968:94, notes 1.48 and l.49): "yoginam gurunirddavyatibhinnarthamatradrk. yoginam apy agamavikalpavyavakTfl)am arthamatradarsanam pratyaksam." The passage is also discussed in Hayes 1984:655-666.

19. The notion of conventionality that is under discussion here is that of conventional truth, which Diimaga takes over from the Buddhist abhidharma tradition of Vasubandhu. See section 3.3.1 above. 20. Pramar;asamuccaya 1:5. The Tibetan reads "du ma'i ngo bo'i chos can nil dbang po las nogs srid rna yin II " Hattori (1968:91) has recorded Prajiiakaragupta's citation of the original Sanskrit: "dharmil)o 'neka-rupasya nendriyat sarvatha gatiJ.!." Please note that the Tibetan translation construes the modifier "sarvatha" as governing the negative "na" and so renders the core of the sentence modally: "rtogs srid rna yin" or "knowledge is impossible." The point is that knowledge of a multi-propertied whole is impossible through the senses. Hattori's translation (1968:27) implies a weaker claim. His translation, "a thing possessing many properties cannot be cognized in all its aspects by the sense," seems to suggest that while sensation can capture some of the aspects of a multipropertied whole, it cannot know the whole exhaustively. But I think the point is clearly that the whole cannot be known at all by the senses, because the notion of a whole is superimposed upon a multiplicity of discrete data of sense. This point is argued in Pramar;asamuccaya 5: 17.

2l. Nyayasiitra l.l.4: "indriyarrhsannikarsotpannarh jiianam avyapadeSyam avyabhicari vyavasayatmakam pratyaksam." 22. Hattori 1968:193. '''khrul ba'i yul nyid kyang srid pa rna yin tel 'khrul ba ni yid kyi 'khrul ba'i yul nyid yin pa'i phyir rol"

Chapter Four

171

23. Hattori 1968:95-97. 24. For other aspects of Diimaga's criticism of the Nyaya doctrine of pratyakiia, see Hattori 1968:36-41 and 121-133, and Oliver 1978. 25. For a good discussion of the history of this debate, see Dravid 1972: 103-130. 26. Pramiil)asamuccaya 1:5. The Tibetan reads "du ma'i ngo bo'i chos can nil dbang po las rtogs srid rna yin I lrang gi rig bya bstan min pal gzugs ni dbang po'i spyod yul yinll" Prajfiakaragupta's citation of the original Sanskrit appears in Hattori (1968:91):

dharmiI)o 'neka-rupasya nendriyat sarvatha gatil:l svasaIiwedya anirdeSyam rupam indriya-gocaral:l 27. For more on the role of causation in inference, see Matilal1968. 28. Lukasiewicz 1951:17. 29. Katsura 1986c:7 attributes this early formulation to Asanga in a text called Shun zhong [un. 30. Both Bochenski (1956:505) and Staal (1967:523) discuss the use of this particle in DharmakIrti, but since they both wrote when Diimaga's system was known only in its barest outlines, neither mentions its use in Dirmaga. 31. The original Sanskrit, quoted by Uddyotakara in Gautama (1967 ed. p. 301), reads "[liilgasya] tattulye sadbhaval:l." Pramiil)asamuccaya 2:5cd is quoted there as "anumane 'tha tattulye sadbhavo nastitasati." Judging from the Tibetan translations, there was an alternative reading wherein "anumeye" occurred in place of "anumane." I find this alternative reading preferable. 32. The word "error" in fact comes from the same Old French root as "erratic," and has as its root sense going astray. The Tibetans fixed upon the expression '" khrul pa" to translate "vyabhiciirin"; the Tibetan word also means to wander out of place, to be wrong, to be deceived, but it also carries with it connotations of being deranged and even mad, and care must be taken by translators from the Tibetan not to read these connotations into the Sanskrit term. 33. The Sanskrit grammarians tended not to give the formal case affixes (vibhakti) names based on their functions, for the semantic functions denoted by the case affixes varied depending upon the surface syntax of the sentences in which the words having the case affixes occurred. Rather than names, the Sanskrit grammarians assigned numbers to the formal case affixes. English-speaking teachers of Sanskrit have unfortunately tended not to follow this example and have instead applied names to the case markers. 1he names usually give are: 1st case = nominative, 2nd = accusative, 3rd = instrumental, 4th = dative, 5th =ablative, 6th = genitive, 7th = locative. While a concern for precision would militate against using these functional names for the cases, the names have become so well established that not to use them entails the risk of hardly being understood. 34. I am paraphrasing Haack 1978:221-222. In her discussion of logical monism and logical pluralism, she says "First, some comments about the conception of correctness which both monism and pluralism require; this conception depends upon a distinction

172

Notes

between system-relative and extra-systemic validity/logical truth; roughly,' a logical system is correct if the formal arguments which are valid in that system correspond to informal arguments which are valid in the extra-systemic sense, and the wffs [well-formed formulas] which are logically true in the system correspond to statements which are logically true in the extra-systemic sense. The monist holds that there is a unique logical system which is correct in this sense, the pluralist that there are several." 35. H.G. Herzberger 1986:66. 36. H.G. Herzberger 1986:66. Indeed, just as Herzberger expected, it is easy to find examples from Buddhist abhidharma of arguments that would serve as counterexamples to the inductive assumption. Here is one: "The faculty of thinking is an external sense faculty, because it is a sense faculty." See below, section 3.1.1 of Chapter six. 37. H.G. Herzberger 1986:67-68. 38. VidyabhU~aIja 1905:218. 39. Gillon 1986 argues that Dharmakirti does not provide a rich enough ontology to support his theory of inference, which includes knowledge secured by causal relations and by counterfactual conditions.

Chapter 5 DiIinaga's nominalism In chapter three we had occasion to examine the interrelationship in the works of Vasubandhu of his inclinations towards phenomenalism and nominalism, and the Buddhist theory of two truths. In the context of that discussion the claim was made that Vasubandhu's work had an influence on Diimaga. One area in which the influence of Vasubandhu upon Diri.naga is especially apparent is in his appreciation of a phenomenalistic view according to which we cannot directly kn()w the external world and according to which it is possible for people to have experiences even in the absence of external stimuli. What was shown to follow from this view was that both words that name complexes and words that name absolute simples can in fact be naming only concepts as opposed to things of the world as they might be independent of our experience. These ideas were explored in various ways by Vasubandhu, as we saw briefly in chapter three. In the present chapter we shall see that Diri.naga also took up these themes, but that he invoked arguments that differed somewhat from those that Vasubandhu had used. We shall begin with a discussion of Diimaga's phenomenalism as discussed in his short and relatively simple treatise called Alambanaparfksa, and then we shall tum to' the elaborate arguments for nominalism that appear in his most mature work, the Pramar;asamuccaya.

S.l The AlambanapanKfjii The Alambanaparfksa is a study in eight verses with a brief prose commentary of the interrelations among several technical terms that Buddhist specialists in abhidharma used to discuss various aspects of awareness. In particular, it follows out a number of ideas from the tradition of Vasubandhu's Abhidharmakosa. The most important of the ideas it

173

174

Diimaga's nominalism

follows out are those of the distinction between conventional truth and rigorous truth, and the definitions of field of operation and support. Once these' technical terms are clearly understood, the argument of the Alambanaparfk$ii is rather straightforward. As we saw in chapter three in the context of the discussion of the Abhidharmakosa, Vasubandhu distinguished the conventionally from the rigorously real, saying "That is conventionally real of which there is no perception when it is broken like a water-jug and when like water it is abstracted 'by the intellect from other things. The rigorously real is otherwise."! According to that view, absolute simples can be regarded as rigorously real, while complex objects made up of collections of simpler parts are real only in the sense that the conventions of human language and thought sanction them as part of the world of our experience. Diimaga takes this distinction between the rigorously and the conventionally real into consideration in his discussion of support (iilambana) and the field of operation(vi$aya) of an act of awareness. The usage of the terms that Diimaga makes is similar to one that Vasubandhu makes in the first chapter of the Abhidharmakosa. The context of the definition of these two terms in the Abhidharmakosa is the discussion of the five external sense faculties and their respective spheres of operation that belong to the set of material properties (nlpaskandha) that make up a person. The five external faculties and their respective fields of operation are all said to be possessed of impact (sapratigha), impact being the characteristic of matter. Vasubandhu distinguishes three types of impact. Tile first is ordinary physical resistance (iivarava-pratighiita) in virtue of which no two material objects can occupy the same space at the same time. The second two are vi$ayapratighiita and iilambana-pratighiita. The first of these two terms, "field impact," refers to the impingement of a particular kind of property, such as colour or noise or aroma, upon the type of sense organ that is sensitive to it. The second term, "support impact," refers to the impact of stimuli upon thought and the emotions. Making this latter distinction invites the question: what is the difference between a property as the field of operation (vi$aya) of awareness and the support (iilambana) of awareness. Vasubandhu's treatment of the topic is quite terse:

as

But what is the difference between the field of operation and the support? Something's field of operation is that in which it has its activity. The support is that which is grasped by thought and mentality.2

Yasomitra (fl. ca. 850 C.E.) glosses the term "activity (kiiritra)" by "puru~akara," which means human activity. So "activity" in this context is interpreted as referring to the eye's activity of seeing colour, the ear's activity of hearing sounds and so forth, and that colour or sound with

Chapter Five

175

respect to which this aCtIvIty takes place is understood as the field of operation of the eye or the ear and so on. But that same colour or sound that serves as the stimulus of a physical organ of sense becomes the support of a thought or an emotion, which grasps the object by means of the sense organ, depending on the sense organ for access to the field of operation in much the same way that a man depends upon a stick to support his weight. Only thoughts and emotions (and other features of what are collectively called "mentality (caitta)") have sUpJ.lorts, whereas both the mind and the sense organs have fields of operation. 3 But in either case what is referred to is that which causes an act of awareness to arise with whatever cognitive shape it assumes. With that much as background, we can now turn to Dinnaga's Alambanaparzksii. After a salutation to all the buddhas and bodhisattvas, Dinnaga begins his examination of the support of awareness as follows: 4 Those who believe that the supports of the types of awareness such as visual awareness are external objects propose either that the supports are atoms, since atoms are the material cause of the awareness, or that the supports are collections of atoms, since awareness occurs with the appearance thereof. About that view, I say first of all Even if atoms are the material cause of the sensory phenomenon, the atoms are not its field of operation, because the phenomenon no more has the appearance of atoms than it has the appearance of the sensefaculty.5

The so-called field of operation is that whose character awareness ascertains, because it is what causes awareness to arise. But atoms, although they are the material cause of awareness, are not that whose character awareness ascertains, in the same way that the sense faculty is not that whose character awareness ascertains. And so, to begin with, atoms are not the supports. As for the collection, although it is that whose appearance awareness has, the phenomenon does not come from that whose appearance'it has. It is correct that the support is whatever object generates the awareness of its own appearance, for it was taught thus as a causal condition of the occurrence [of awareness). But the collection is not what generates the awareness of its own appearance,

because, like the second moon, it does not really exist.

In seeing a second moon owing to a deficient sense faculty, although there is the appearance of a second moon, that is not its field of

176

Diilnaga's nominalism operation. Similarly, a collection [of atoms] although it is the material cause of awareness, is not the support, because it does not really exist.

Thus neither external Qbject can be the field of operation of awareness. The support is neither the external object called the atoms nor the external object called the collection of atoms, because each of these is deficient in some respect.

Tl}.is passage is clear enough to be virtually self-explanatory and requires very little elucidation, but it might nevertheless be instructive to try to translate the argument into a slightly different terminology. We can start by availing ourselves of the useful distinction between an object as perceived in awareness and an object as it is in itself independent of anyone's awareness of it. That which is perceived in awareness is what we called in chapter four the sensum or sense datum, which occurs as a coloured shape, a sound, an aroma or some other property as received through the channel of one of the five physical senses or through the channel of the intellect in the form of a thought. Presumably there is some object outside the physical body of the being who is aware that somehow impinges upon the physical body of that being, and when this impact occurs, there arises an act of awareness of a coloured shape or a sound. What is not clear is to what extent the nature of the resulting sensum resembles the nature of the external object that served as a stimulus to the sensum's arising. It is towards the end 01 clarifying that issue that Diilnaga embarks on his investigation of the support of awareness. He is willing to entertain the view· that what really exists in the external world is a multitude of various kinds of atoms, all of which are ultimately tiny and hence have the same dimensions. But this is very far from what any being actually senses. The characteristics that sensa have are such things as being jug-shaped, house-shaped and cat-shaped patches of colour, but no atom has the characteristic of being a jug-shaped or cat-shaped patch of colour. An atom has the characteristic of being atom-shaped, whatever shape that may be; if atoms are without dimension, it may mean having no shape and no colour at all. So there is a considerable discrepancy between the nature of the presumed stimulus of an act of awareness and the object as perceived in awareness. This being so, we cannot really say that the object of which we are aware is atomic in nature. When, for example, a cat opens her eyes and looks around, it is not earth atoms that she is aware of, but mice and other items that occasion interest or alarm or indifference. But a mouse is composed of atoms, and as such it is a whole made up of parts, and as such a mouse is not, according to Buddhist abhidharma, a rigorously real item. At best it is a

Chapter Five

177

conventionally real item. But what this means is that sentient beings are never aware, at least not through sensation, of what is rigorously real. But since it is only what 'is rigorously real that can in the final analysis be said to have any actual effect on the world in general and in particular upon that part of the world that is conscious, it follows that what actually causes a sensation to arise is never that about which we have awareness. There is, then, an unbridgeable gap between reality and appearance. The real nature of the world is forever hidden from us. The only things that are not opaque to us are the sensa themselves, just as they appear as phenomena in awareness. Nothing but a sensum has the shape, or whatever other sensible characteristic one senses, that a sensum has. As we noted above, when one senses a jug-shaped patch of colour, it is the sensum alone that is jugshaped, for the atoms are not in the shape of jugs, and the collection of atoms is merely a conceptual fiction that cannot be said to have any actual characteristics at all. This leads Diimaga to conclude in his Alambanaparzksa: Difference in shape occurs only in what is conventionally real, but it does not exist in the atoms themselves [which alone are real in the rigorous sense]. The water-jug and so forth are conventionally real because if the atoms are taken away, the awareness of their form is destroyed. In the case of what is rigorously real, such as colour and so forth, even when one has taken away what is connected with it, there is no removal of the awareness of the colour itself. Therefore, it follows that the field of operation of the sensory awareness is not an external object. Rather, the object is that internal object of awareness which appears to be external. The support is just that which occurs internally, appearing to be external in the absence of an external object, because it has the characteristic of awareness itself, and because it is the causal condition of awareness. The support is just that which occurs internally, for, given that the internal object of awareness appears to be the object and that awareness arises from it, it has the two characteristics [of the field of operation of awareness as outlined above].

Although his arguments are different from Vasubandhu's, Dirinaga's conclusion is a phenomenalism quite similar to that which appears in the conclusion of the Vijfiaptimatratasiddhi. And as was the case in Vasubandhu's system of thought, Dirinaga's phenomenalism resonates

178

Dinnaga's nominalism

harmoniously with both a skepticism of the sort that was discussed in the preceding chapter and a nominalism of the sort that appears in the PramalJasamuccaya. To see the form of nominalism that developed in Dinnaga's system of thought we must turn our attention to the fifth chapter of the Pramauasamuccaya.

5.2 The context of the discussion of nominalism in the Pramii1)asamuccaya In his opening comments to the PramalJasamuccaya, Dinnaga states that his purpose in writing the work is to remove certain misconceptions that people have about the acquisition of knowledge. Among these misconceptions, he says, is that there are more than two standards to be used in assessing whether a belief qualifies as knowledge. Dinnaga himself recognizes only two such standards, namely, the test of direct sensation and the canons of logic. Any belief that is not founded upon what is directly sensed or upon a line of reasoning that stays within the bounds of careful reasoning must be discarded as an unwarranted belief. This matter of whether there are means of acquiring knowledge outside sensation and well-founded reasoning is the principal topic of the fifth chapter of the PramalJasamuccaya, where it is stated that the purpose of the chapter is to show that all other means of acquiring knowledge that are named by other philosophers can be reduced to forms of reasoning. What made this doctrine particularly controversial was that it reduced the authority of scripture (agama) to the authority of reason. That is, scriptures could according to Dinnaga be recognized as authoritative only to the extent that they could meet the strict tests of logical soundness. This doctrine, which was not new with Dinnaga but was in fact a common feature of Buddhist philosophy, was particularly repugnant to orthodox Brahmanical philosophers. In order to gain a better perspective on some of the issues involved, it is worth taking a brief look at how Bhartrhari, a contemporary of Dinnaga in opposition to whom Dinnaga formed many of his ideas, defended the doctrine of the irreducible authority of scriptures. 5.2.1 Scripture as a form of inferential sign Bhartrhmi's contention that scriptures cannot be reduced to inference is based on two separate arguments, namely, that scriptures have a unique subject matter and that inference, unlike scripture, is -fallible. To

Chapter Five

179

demonstrate that scriptures have a unique subject matter, Bharq-hari appeals to the observation that it is only from scriptures that we can acquire knowledge about such things as religious duty (dharma). In the first chapter of the Viikyapadfya, entitled "BrahmakiifJda" or "AgamakiifJda," Bhartrhari says And without sacred tradition, religious duty is not fixed by means of reason. Even the knowledge of the Seers was preceded by sacred tradition. No one can refute by reason the uninterruptedly fixed modes of duty, for they are well known to people. 6

. Furthermore, it is only scripture that gives man knowledge of moral matters. "Men, including outcastes," he writes, "have little use for academic treatises in the two matters of what is right and what is wrong."7 In this passage, the expression "academic treatises" refers to manmade texts in contrast to the eternal scriptures revealed to man through the agency of a Seer (~$i) or prophet. And finally, certain extraordinary powers belonging to classes of beings that are not part of the experience of ordinary people, and such doctrines as the law of karmic fruition and rebirth come to be known to humanity ultimately only through the scriptures: The special powers of the ancestors and of the Rakshasas and of the Pisacas, which derive from their deeds [in former lives], are outside the realm of sensation and inference both, yet they are established facts ... .Inference does not repudiate the statements of those who with the Seer's eye see entities that are beyond the senses and beyond the intellect. s

Since religious, moral and metaphysical truths can be neither established nor repudiated by sensation or reasoning, claims Bhartrhari, knowledge about them can be acquired only from some source that is independent of sensation and reasoning. 5.2.2 Fallibility in inference and scripture Bhartrhari's second argument, that reasoning is a fallible source of information, appeals to the observation that an inference about a particular matter is reliable only to the extent that the knowledge about the thing is complete. A person with less experience in a particular field may make a valid inference, but the conclusions of that inference may be subject to revision by a person with greater expertise in the field. But even the expert's knowledge is incomplete in the sense that the possibility always remains that a thing whose behaviour is uniform in every known

180

Diimaga's nominalism

circumstance may yet behave differently in an as yet ,unknown circumstance. Bhartrhari writes: It is extremely difficult to gain a good understanding through inference of the capacities of things, capacities that vary in condition, place and time. The capacity of a substance of which the capacities in a variety of utilities are well known may be hindered in conjunction with a particular substance. A thing, even though it be carefully reasoned out by a skilled judge9 is shown to be quite otherwise by another who is more experienced!

While an inference is always liable to be overthrown by further experience, scriptures, which come down to man from beings of unsurpassable expertise, cannot be thus overthrown. Viikyapadfya 1:37 refers to beings of such clarity of mind that they can see the past and future directly, an ability that would minimize surprises and thus confer an undeniable advantage over even the most expert scholar confined to ordinary modes of learning. And so it is by appealing to the existence of omniscient beings who have a special means of avoiding the pitfalls of inductive reasoning that Bhartrhari argues that scriptures, an infallible source of information, cannot be reduced to inference, a fallible source. DiIinaga does not deal directly with the passage of Bhartrhari that we have just outlined, but he does deal with the key issues involved in that passage. First of all, that scriptures contain statements the truth of which cannot be ascertained by any other means than through the scriptures themselves is not, in DiIinaga's view, a reason for concluding that those scriptural statements are true. Rather, statements the truth of which cannot be ascertained through sensation or reasoning are statements the truth of which cannot be ascertained at all. In Pramii1Jasamuccaya 2:5ab and his own commentary on it, for example, DiIinaga cites the Sarhkhya doctrine that all mental and physical phenomena are transformations of a primordial substance (pradhiina) as a doctrine the truth of which is unknowable. Presumably his reason for declaring primordial substance unknowable is that such a substance is in principle impossible to experience in its primordial form, but one can never in DiIinaga's view infer the general form of anything if one has not previously experienced a particular instance of it. Thus in general DiIinaga's view of those religious, moral and metaphysical doctrines that can neither be established nor repudiated by direct experience or reasoning would most likely be that it is a mistake to say that knowledge of them can be acquired only through scriptures. At best we can say that information about them is acquired from scriptures, but about the accuracy of that information there can be no certainty. A distinction is to be made between a source of information and' a source of knowledge, and scriptures may be the former without being the latter.

Chapter Five

181

The second of the two claims of Bhartrhari that was outlined above was that inference is a fallible process. Its fallibility is demonstrated by the fact that two reasoners may arrive at contradictory conclusions. This event· of two reasoners arriving at contradictory conclusions is possible only if at least one of the reasoners avails himself of· an unsound argument, since two lines of reasoning both of which are merely valid can lead to contradictory conclusions, but two lines of reasoning that are sound cannot. Diimaga does not dispute Bhartrhari's claim that not all reasoning yields knowledge. Where he disagrees with Bhartrhari is in his claim that if scriptures yield knowledge about that which is not present to the senses, then they do so because the statements in them can be formulated as valid arguments the truth of whose premisses is known. But if scriptural assertions fail to be capable of reformulation as sound. arguments, then those scriptural assertions do not yield knowledge. Scripture, in other words, is fallible to exactly the extent to which reasoning is fallible. Hence scripture, and indeed all other forms of verbal communication, are to be subsumed under the rubric of reasoning in general, and verbal communication that yields knowledge is subsumed under the rubric of sound reasoning. The declared purpose of the fifth chapter of the PramalJasamuccaya is, as we have already noted, to clear up the misconception that there exists any substantial difference between the inferential process and the process of acquiring knowledge through linguistic signs or symbols. Since in DiIinaga's view verbal communication is but a species of reasoning, we find that many of the issues raised in his discussion on reasoning are reflected or paralleled in his discussion of verbal communication. Therefore, before examining DiIinaga's ideas on verbal communication, which is the special topic of .the fifth chapter of the PramalJasamuccaya, it will be profitable to review some of the key points of his discussion on sound reasoning as they are treated in the second chapter of that work. As was noted in the preceding chapter, reasoning for DiIinaga is a process of attributing a property to a subject on the basis of a piece of evidence (hetu) or inferential sign (linga) that is itself a known property of the subject in question. In order to progress from the knowledge that the subject in question is a locus of the property being used as evidence to the knowledge that that subject is also a locus of the inferable property, it is also necessary to know one further thing, namely, that the evidence is restricted to the inferable property. To say that one property P is restricted to a second property R is to say that every locus that is or has an instance of P is also a locus that is or has an instance of R. DiIinaga has a variety of

182

Diimaga's nominalism

ways of expressing the fact that P is restricted to R, but all these,formulae express the same state of affairs, namely, that the set of loci that are the extension of property P is a subset of the set of loci that are the extension of property R. . As was also noted in the preceding chapter, in order for an inference to produce a conclusion that is certainly true, it is necessary that one know that the property being used as evidence is restricted to the property being attributed on the basis of the evidence to the subject of the inference. This means that one must know that every instance of the property being used as evidence occurs in the presence of, or shares a locus with, an instance of the attributed property. The problem that this stipulation raises, as Bhartrhari suggested in his discussion of the fallibility of inference, is that knowledge of this magnitude can be had only by one who is omniscient in the strictest sense of that term, for only an omniscient being could know every past, present and future locus of the attributed property. Therefore, since in the realm of ordinary mortals one cannot be certain of the restriction of the evidence to the attributed property, one cannot be certain of the truth of the conclusion of an inference based upon that restriction. As has also been claimed above, Diimaga does not dispute Bhartrhari's claims concerning the fallibility of inference. The claim that no inference based on a statement of one property's restriction to another is capable of producing a certainly true conclusion is not, of course, inconsistent with the claim that certainty in inference is possible only if one knows that the one property is restricted to another. This latter claim states the minimum condition for certainty in inference, and that no piece of reasoning meets that minimum condition does not make the statement of the minimum any less valid. In matters of practical reasoning, however, DiIinaga accepted the principle that the non-observation of a counterexample to a universal proposition serves as justification for believing that universal proposition. In other words, one is entitled to believe that property P is restricted to a property R so long as one has not observed an instance of P in the absence' of an instance of R. DiIinaga deals with this issue in the fifth chapter of the Pramiir;asamuccaya in his own commentary to verse 35. Here the topic is specifically linguistic signs, but it is clear that what he says there is applicable to the question of signs in general. He writes: Association and dissociation are the two ways that a verbal symbol expresses its object. They consist respectively in applying to what is similar and in not applying to what is dissimilar. It is not necessary to say that a verbal symbol applies to every instance of what is similar, because in some cases it is not possible to express an extension tpat is unlimited. But it is possible to say that it does not occur in the dissimilar--although it too is unlimited--simply on the basis of its not

Chapter Five

183

being observed to apply to any dissimilar instance. For this reason, because a term is not seen to apply to anything other than that to which it is related, its expression of its own object is said to be a negative inference.

5.2.3 The question of universals As the matter has been stated in the preceding chapter and up to ' this point in the current chapter, it would seem natural to interpret the properties of which we have been speaking as universals. That is, it would seem natural to interpret "x is a locus of property P" as equivalent to "x is a locus of the universal P." There are, however, passages in Diimaga's work that disallow such an interpretation. At PramalJ.asamuccaya 2: 16, for example, Ditmaga states his view that universals ani non-existent, or, if existent, at least unknowable. Before we can go any further in the discussion of inferential knowledge, it is necessary to recapitulate Diilnaga's reasons for stating that universals are unknowable, and then we must see how "x is a locus of property P" is to be understood in his system thought. The concept of the universal (jati) towards which Diimaga addresses his discussion is that of an entity that is itself indivisible and that resides in a plurality of individuals. This notion of universals as entities that are single while residing in a plurality of particulars is at least as old as Patafijali, who alludes to these features in the Paspasahnika section of the VyakaralJ.a-mahtibhtisya.1O Diilnaga argues that such an entity is logically impossible on the grounds that the two predicates "indivisible" and "resident in a plurality of individuals" are incompatible. l1 The full line of reasoning goes as follows. A universal's residence in an individual must be either complete or partial, that is, either the entire universal resides in an individual or only part of it does. If a universal U resides in its entirety in given individual "1' then it does not reside at all in individuals "2' "3' u 4 , .•• ,u,. and thus fails to be resident in a plurality of individuals. If on the other hand the universal is conceived as residing only partially in each of its individual instances, then it loses its indivisibility, for it then has as many internal divisions as there are individuals in which it supposedly resides. Independent of the above argument against the possibility of universals as indivisible entities that reside in a plurality of individuals, there is another argument to the conclusion that universals are unknowable. 12 The concept of universal here appears to be that of an organizing principle such as, perhaps, resemblance, by virtue of which a

184

Diilnaga's nominalism

plurality of individuals are collected together into a class. Even if there is such a principle, knowing it is impossible without knowing each individual in the class determined by that principle, for until we know each individual in a class we cannot know what they have in common that qualifies them all for membership in that class. But to know each past, present and future individual in a class is clearly impossible for anyone who is spatially and temporally finite. Thus it would be impossible to know that x is a locus of universal P--or, as it was more usually expressed by Sanskrit philosophers, "x is qualified by the universal P"--since such knowledge would require knowing both x and the universal P. Now the question arises whether there is an interpretation of an expression of the type "x is a locus of the property P" whereby it expresses only that which is knowable. Diilnaga thinks that there is an affirmative answer to this question. In order to make some sense of his answer it is helpful to realize that, while we may not know the organizing principle by virtue of which a plurality of individuals are collected together, we can know the very fact that they are collected together. Thus we may begin with the knowledge that there is a set or collection P made up of a number of individuals PI' P2' P3 and so on. To make some sense of the expression "the property P" or "the universal P" we must link it up somehow with the collection P. The expression "the collection P" can in tum be seen as another way of saying "all p's." Now it is obviously not the case that "the universal P" is simply an alternative way of saying "all p's," for then "x is a locus of the universal P" would be rewritten as "x is a locus of all p's," which is clearly not what we wish to say. What we do wish to say is "x is a locus of some p." This latter expression is easily arrived at by denying the expression "x is a locus of no p." So whenever we encounter an expression of the form "x is a locus of the universal P," we are to interpret it as saying "x is a locus of the deniability of the absence of every particular p" or "x is qualified by the exclusion (apoha) of the absence of every p." The advantage that Diilnaga seems to have thought is inherent in the above interpretation is that whereas the truth of "x is qualified by the universal P" cannot be known because of the unknowability of the universal P construed as an organizing principle, "x is qualified by the exclusion of the absence of every p" can be known by knowing only one particular, for knowing that one particular p is present at x is sufficient to falsify the proposition that every particular p is absent from x.

Chapter Five

185

5.2.4 Anyapoha as a substitute for universals Let us now apply the above apoha interpretation to a standard line of reasoning, whereby from the observation of a particular case of smoke at a place it is inferred that there is a fire at that place. A line of reasoning of this sort might be expressed as follows: 1. There is smoke at x. 2. No locus that has smoke lacks fire. 3. Therefore, there is fire at x. Statement (1) in the above set of statements, when rewritten in compliance with the apoha theory is to be understood as follows: (la) Locus x is qualified by the deniability of the absence of every particular body of smoke. The absence of every pa.'1icular body of smoke at x is made deniable by the presence of at least one particular body of smoke at x. This if a particular body of smoke, which is here serving as the evidence on the basis of which fire is attributed to the locus x, is known to be at x, the first of Diimaga's three characteristics of legitimate evidence is in place, for this characteristic can be expressed as "The evidence must be a property of the subject to which the attributed property is attributed." The apoha interpretation of statement (2), which states the restriction of the property that is used as evidence to the attributed property is: (2a) No locus that is qualified by the deniability of the absence of every particular smoke is qualified by the absence of every particular fire. If this statement is true, then the third of Ditmaga's three characteristics of proper evidence is in place, for the third characteristic can be expressed "The evidence must not occur in any locus that lacks the property that is being attributed to the subject of the reasoning at hand." If the first and third characteristics are in place, we can be sure of the truth of the statement (3), which is interpreted under the apoha theory as follows:

(3a) Therefore, locus x is qualified by the deniability of the absence of every particular fire.

186

Diilllaga's nominalism

As was the case above, the absence of every particular fire at x is made deniable by the presence of at least one particular fire at x. As I have already suggested, the purpose of this apoha interpretation is to emphasize that the knowledge gained through the inferential process is not that· there is fire at x, but rather that there is a partic.ular fire at x. That is, we are not to conclude that x is qualified by the fire universal, but rather we are to conclude that x is qualified by the presence of at least one particular fire, which can also be expressed as the deniability of the absence of every particular fire. Knowing of x only that it is qualified by the deniability of the absence of every particular fire does not, of course, include knowing any of the particular features, such as the colour or shape or temperature or the fuel that serves as a cause, of the particular fire that is at x. In other words, we cannot know the identity of the particular fire at x. Rather, of the particular fire that is discovered through reasoning to be at x, only the general features are knowable, the general features of fire being just those features the absence of all of which can be denied of every particular fire. 13 For example, since the deniability of the absence of every particular fire is restricted to the deniability of the absence of every particular substance, and this is tum is restricted to the deniability of the absence of every particular reality, we can safely conclude that the particular fire at x is a particular substance and a particular reality. The general knowledge gained through the inferential process, then, is general in the sense that we know only that some particular is present at a given locus, but we do not know which particular is present there. The knowledge is not, according to Diilllaga, general in the sense of dealing with real universal properties. The reason for emphasizing the point that the general knowledge to which inference leads is general knowledge of the form "some particular is present there" is that particulars are knowable while universals are not. Diilllaga's programme of replacing the concept of a universal U with the deniability of the absence of all particular u's raises a number of problems. One key problem that confronts his interpretation of universals is this: If one does not know what each u must have in order to be counted within the range of the expression "all u's," then one cannot determine whether a particular is qualified by the absence of all u's or whether by the deniability of the absence of all u's. But if one does know what each u must have to be counted within the range of "all u's," then one knows exactly that which is ordinarily called a universal. And therefore, it could be argued, the success of the programme of dispensing with universals is only apparent.

Chapter Five

187

Diilnaga's answer to the above objection is to say, in effect, that the universal property that is shared by a group of particulars and that serves as the basis for grouping those particulars is simply the fact that all those particulars are members of the set of things to which a specified. word is applicable. That by which I know of a fire that it is a fire is simply the fact that I readily apply the word "fire" to it. The problem now becomes one of how we know of a given particular that a specified word is applicable to it, and Diilnaga's answer to this seems to be that our knowledge of the applicability of words to things is not necessarily reducible to knowledge about any other sort of thing than our own linguistic habits. In other words, without knowing exactly what it is that a thing must have in order for me to apply the word "fire" to it, I can still have a sense of confidence in the correctness of my decision about nearly any particular whether that particular belongs or does not belong to the set of things to which the word "fire" is applicable. And in those cases in which I do not have such confidence, the lack of confidence stems from my lack of clarity about how a tenn should be used rather than a lack of clear understanding of the thing to which the tenn is to be applied. I may not know, for example, whether a particular item is a monkey-wrench, but that lack of knowledge stems from a lack of decisiveness in my own mind about the proper use of the word "monkey-wrench." This whole disc;ussion of words and their applicability is the principal subject of discussion in the fifth chapter of Pramiil;asamuccaya. The chapter comprises fifty verses, not counting those that Diilnaga quotes from other authors, that deal with a variety of topics all of which have some connection with language. Beginning with the contention that a linguistic sign functions in the same way as. an inferential sign (Linga) to produce general knowledge, Diilnaga then discusses the content of that general knowledge so produced. What this amounts. to is a discussion of what it is that an individual linguistic sign such as a word expresses. This discussion takes up the first thirteen verses, or about one quarter of the chapter. Following this is a discussion of co-reference (siimiiniidhikarafJya), the principle that two or more words with different intensions may apply to the same object or set of objects, and the qualification relation (visesafJavise~abhiiva) whereby one linguistic expressions qualifies or na.rrows the scope of others. This discussion as a whole deals with the problem of how strings of symbols function together to produce knowledge of complex states of affairs, and it takes up the next twelve verses. The next thirteen verses after that deal with the principle that a word's applicability to a particular object precludes the applicability of certain other words to that same particular. The key problem here is to determine which words are precluded and which are not precluded once a

188

Diimaga's nomimi.lism

given word is applied to an object or set of objects. The final quarter of the chapter· deals 'with a number of miscellaneous questions such as the meanings of sentences, the meaning. of singular terms, and the learning of the meanings of words. Since it is in the first three quarters of the chapter that Diimaga's most original ideas emerge,I shall concentrate on what is contained there and give onlY'a brief account of the concluding section of the fifth chapter.

5.2.5 TlJ,e nature of information conveyed by language In the opening 'passage of the fifth chapter of the Prama/')asamuccaya, called Apohapariccheda, Diimaga states that verbal communication is in principle no different from inference. It is important to bear in mind that this statement is made with reference to what knowledge a person acquires upon hearing a symbol used. The claim is that a verbal symbol tells its hearer something about the object to which it is applied in the same way that a piece of evidence tells its observer something about the object in which it occurs as a property, and moreover the nature of what each of these two things tells about its respective object is essentially the same in each case. What Dinnaga saw as the parallelism between a linguistic symbol and a property used as evidence for a second property requires some elucidation, for it is not entirely self-evident. We may begin with some fairly obvious observations of what sorts of things are required in order for a person to acquire knowledge as a result of hearing a verbal symbol. First, the hearer of the symbol must know the conventions that have been established with respect to the use of that symbol to stand for certain things. Second, the hearer must know that the speaker of the symbol that the hearer hears is using the symbol in confoTIl1ity with those conventions; for clearly if the speaker is deliberately or accidentally using a symbol for something other than that for which the symbol conventionally stands, the hearer will not gain true information through hearing it. What is involved in knowing the conventions governing the use of the symbol need be no more than knowing that the symbol is applicable to some things and is inapplicable to others, and of course the hearer should have some knowledge of at least some of the specific things to which it is not applicable. These requirements that have just been enumerated for a symbol's competence to yield knowledge to the hearer of the symbol can be restated in a way that bears remarkable resemblance to the requirements stated by Dinnaga for a property's competence to yield knowledge to the observer of that property. In order for a word to be

Chapter Five

189

competent to yield knowledge, it must have the following three characteristics: 1. The word must be applicable to the subject of discpurse. 2. The word must be applicable to objects other than the subject of discourse that have that which is to be learned through the word. 3. The word must be restricted in application to that which is to be learned through it.

Concerning the first of the above three characteristics, a linguistic sign's being applicable to a particular thing is not unlike a property's occurrence in a particular thing, In fact, in Sanskrit the same verb, "vartate," is most often used to express both a property's occurrence in an object and a word's applying to an object. Thus we find such expressions as "utpalatvam tasmin dravye vartate" for "The property of being a lotus occurs in that thing," and "utpalasabdas tasmin dravye vartate" for "The word 'lotus' is applicable to that thing." Thus a symbol's applicability to a particular, the subject of discourse, makes known or expresses some fact about that particular in much the same way that a property qua evidence by occurring in a particular, the subject of inference, makes know some fact about that particular. In each case, knowledge about this further fact about the particular is contingent on knowledge of the sign's restriction to the learned fact, that is, the symbol's restriction in application to that fact or the evidence's restriction in occurrence with that fact. Incidentally, it is clear that Diimiiga's discussion of linguistic signs is primarily about audible symbols in the form of spoken language. It would not, however, be difficult to extend the discussion to visible symbols such as printed words as well. The principle of an audible symbol's applying to an object is in no important respect different from a printed symbol's application to an object. Concerning the second of the characteristics described above, that a symbol in order to convey knowledge must be applicable to objects other than the subject of discourse, this requirement would at first glance appear to eliminate singular terms from the set of symbols that can yield knowledge when heard. This, however, is not the case, since individuals, which are the referents of singular terms, are regarded by Diimiiga to be the synthesis of a mUltiplicity of cognitions and hence are treated as classes rather than as particulars. 14 A particular for Diriniiga is the content of exactly one cognitive act; it is that which gives a single act of cognition its identity. When, however, the contents of a multiplicity of individual cognitive acts are regarded--and this is essentially a subjective judgement-to have sufficient in common to be regarded as all being about the same thing, then this "same thing" they have in common may be given a name.

190

Diri.naga's nominalism

That name may be something like "man" or it may be something like "Socrates" or "Siddhiirtha." The only difference in our applying "Siddhiirtha" and our applying "man" to the contents of a multiplicity of cognitive acts is a difference of degree. The cognitions to the contents of which we apply "Siddhartha" appear to have more in common than those to which we can apply only "man." Thus what this second requirement eliminates is not singular terms, but words for which there is no established convention at all. I might, for example, apply the term "rejaxepoid" to the figure I am now imagining or seeing, and as long as I keep it to myself what the figure I am now imagining or seeing looks like~ no one who hears me speaking of a rejaxepoid will know to what objects the word "rejaxepoid" is applicable and to what objects it is inapplicable, and hence the word will yield no new knowledge to anyone who hears it. It is idiosyncratic words such as this that are eliminated by the second requirement mentioned above. Concerning the third of the characteristics mentioned above, that a symbol in order to be capable of conveying knowledge of a property must be restricted to that property, Diri.naga discusses as PramalJasamuccaya 5:35 the nature of our awareness of conventions restricting the use of terms. His view is consistent with the view traditionally put forth by the Sanskrit grammarians. According to the account of the traditional grammarians, knowledge of what a symbol applies to or what it means is acquired through knowledge of association (anvaya) and dissociation (vyatireka). In the grammatical context, association refers to the applicability of a symbol to an object when that object is endowed with a certain property, and dissociation refers to the inapplicability of that symbol when the property in question is absent. 15 Diri.naga follows this traditional account by arguing that anvaya, the joint presence of a symbol and a property, or in other words the applicability of a symbol S to an object x when a property R is present in x, and vyatireka, the joint absence of a symbol and a property, play interdependent roles in the hearer's inference from symbol S to property R. The gist of his argument is as follows. Suppose that a hearer of a symbol S knows about the applicability of that symbol no more than that it is applicable to a certain set of objects M and that all members of M have in common several properties, which we will call P, Q and R. Now suppose further that the extension of property R is a subset of the extension of Q and that the extension of Q is in turn a subset of the extension of P. Now presumably when a symbol S is applied to each of the members of M, the symbol is intended to express only one of the properties P, Q and R. But of course if the hearer only knows that S is

Chapter Five

191

applicable to each object in M and that each object in M has each of the three properties P, Q and R, then he lacks sufficient information on which to base a judgement on which of the properties is named by S. Moreover, if it tums out that in fact S is intended to express just R, then no matter how many times the hearer is shown that the symbol S applies to an Rpossessing object, he will still not have a means of knowing whether S is intended to express P or Q or R, because every R-possessing object also possesses properties P and Q. In the situation we have just been considering, the hearer of S could know at most that the symbol S applies to some R-possessors and to some Q-possessors and to some P-possessors. But this knowledge is clearly not sufficient to enable one to infer, on the basis of observing some future application of S to a heretofore unknown object y, that y possesses property R. What enables the hearer of S to infer the presence of R in any object to which S is applicable is not the knowledge that S applies to some or all R-possessors. Rather, it is knowledge that S is restricted to them, that is, that S applies only to Rpossessors, that enables the inference to take place in a reliable way. In other words, it is necessary to know no more and no less than that the symbol S is not applicable to any object x such that x is not an instance of property R. Therefore, claims Diri.naga, it is neither association alone nor dissociation alone that provides the grounds for a hearer's forming judgements about things from hearing a word or words applied to those things, but rather the combination of association and dissociation. Up to this point we have seen that some similarity can be observed in the manner in which a linguistic sign or conventional symbol tells its hearer something about an object and the manner in which an inferential sign or piece of evidence tells its observer something about an object. What remains to be seen is how the nature of what a symbol makes known or expresses is similar to the nature of what a piece of evidence makes known. As was noted above, a piece of evidence, which we may here call property H, makes known the fact that the object in which it occurs is not among the objects that lack a further property, which we may call property R, to which the evidence is restricted. But symbols, which function in the same manner as evidence, may be seen as simply a special kind of evidence that, when restricted in application to a property, make that property known. Thus if we can say (1) If property H is restricted in occurrence to property R, then if property P occurs at locus x, then property R occurs at locus x,

then we can also say

DiIi.naga's nominalism

192

(2) If symbol S is restricted in application to property R, then if symbol S applies to locus x, then property R occurs at locus x.

It was also noted above that, while it is convenient to talk of properties, in DiIi.naga's view it is possible to eliminate such talk by replacing phrases that refer to the occurrence of a property R with phrases that refer to the deniability of the absence of every particular r. In this context the symbol "R" stands for a universal or repeatable attribute, and the symbol "r" stands for a particular instance of that attribute. This aforementioned elimination supposedly enables us to speak only in terms of particulars without having to avail ourselves of talk about universals. For the sake of completeness it should be added that when we speak of a symbol, such as the word "fIre," we are often referring to that real or imagined quality that every instance of the symbol has in common; thus "fire", "fire", and "fire" may be regarded as three separate instances or symbol tokens of the one symbol class "FIRE", which is conventionally written in upper case letters to emphasize that it refers to a symbol class and not a symbol tokea. Just as we can eliminate "property R" in favour of an expression referring only to particular r's, we can eliminate "symbol S" in favour of an expression referring only to symbol tokens. We can now write statement (2) above as follows: (2a) If not every symbol token s is inapplicable to x only if not every particular r is absent at x, then if any symbol token s is applicable to locus x, then it is not the case that no particular r occurs at locus

x.

Thus a restricted symbol makes known exactly what a restricted piece of evidence makes known: the deniability of a universal negative proposition concerning the occurrence of particular instances of the property to which the symbol is restricted. If anything has been demonstrated so far it is only that if one is persistent then one can replace some discourse that seems to refer to universals with somewhat more cumbersome discourse that seems to refer only to particulars. What has yet to be demonstrated is why anyone should avail himself of a cumbersome mode of discourse when a much simpler one is available that seems to serve all the same purposes, except the dubious purpose of eliminating all mention of universals. In fact, it has yet to be demonstrated that the apoha interpretation is even capable of replacing the talk of universals and qualities ih all contexts,in which speaking of universals and qualities has proven so useful. Most of the

Chapter Five

193

material in verses 2 through 25 of the fifth chapter of PramafJasamuccaya deals at least indirectly with the above issues. The general question of why one should adopt an apoha interpretation, or to put it another way, why one should accept that words express exclusions of courfterextensions rather than properties, is answered by showing that there is no other suitable candidate for what it is that linguistic signs express. So the discussion now turns to why it is that words cannot be regarded as expressing individuals, universals, the relations of universals to particulars, or particulars qua instances of universals. 5.2.6 The meaning of individual words As has already been said, the question of what a word or expression means is for Diimaga a question of what the hearer of that word can come to know when he hears the word used, and this in tum is a function of the word's restriction in applicability to a certain domain (visaya) of particulars. Given this, it is easy to see why he argues that a word cannot express any proper subset (cf. bheda) of the set of particulars to which it is applicable; it would be nonsense to say that a word is applicable only to some but not to all of the things to which it is applicable. But even if a word cannot by itself pick out or yield knowledge of just one particular or any number of particulars less than the total number of particulars to which it is applicable, why can we not say that the word picks out the totality of particulars to which it is applicable and thus makes that totality known? That would be tantamount to saying that when I hear, for example, the word "woman," I gain knowledge of the entire set of beings to which that word is applicable, namely, to each and every particular woman in the present as well as in the past and future. It is clearly not the case thai words have such far-reaching capacities to yield knowledge. A word, then, cannot express either all or only some of the particulars to which it is applicable. 16 If, as the above argument claimed to show, a linguistic sign does not express its extension, the set of particulars to which it is applicable, it would seem natural to conclude that it expresses its illtension, the property in virtue of which all members of the set of particulars to which the word is applicable belong to that set. Thus a word such as "cat," while it cannot make all cats known to one who hears the word, can make cathood or felinity known, for the word "cat" is restricted ill application to just those situations in which felinity is a factor. The main reason that Diimaga finds this account unacceptable is that, although it may account for what individual symbols express, it does not adequately explain how a symbol functions when it is part of a string of symbols. In fact, Diri.naga will

194

Dirinaga's nominalism

ultimately deny that even isolated words can express intensions if by intensions we understand real universals. We shall come to this argument further on, but for the time being let it be granted that words in isolation express universals. Even granting this much, it is arBued that symbols as part of symbol strings do not express their intensions. Consider for example a string of symbols wherein a predication is made. Such a string might be a grammatically complete sentence such as "A cat is an animal" or an appositional juxtaposition of words such as "the animal, cat" or "the cat, an animal" that could be part of a longer string such as "my favourite domesticated animal, the cat." In either case the string represents what may be called a logically complete sentence, or a simple predication. Now if in a string of this sort we understood each symbol to be expressing a universal, that is, if "cat" expresses felinity and "animal" expresses animality, how are we to understand the relation between these two universals when "cat" and "animal" are in apposition? It should be obvious that the relation is not one of identity, for it is not the case that the set of characteristics that a thing must have in order to have felinity is exactly the set of characteristics that a thing must have in order to have animality. But rather the relation between felinity and animality is a whole-part relation in the sense that the necessary conditions of animality are only part of the necessary conditions for felinity. That Mungojerrie is an animal is part but not all of what is expressed by "Mungojerrie is a cat." It so happens, argues Dirinaga, that there are a number of ways by which whole-part relations are reflected in strings of symbols. If we wish to convey the fact that a neck is part of a giraffe, for example, we make use of such expressions as "the neck of the giraffe," "the giraffe's neck, "the giraffe has a neck" and so forth. In each of these symbol strings the whole-part relation is explicitly signaled by some part of the string other than the symbol for the whole and the symbol for the part. We do not, in other words, express a whole-part relation by simply juxtaposing the symbols for the relata as in *"the neck the giraffe" or *"the giraffe the neck." We should, then, in a string of symbols that reflects the whole-part relation between felinity and animality, expect to find some part of that string that signals this relation explicitly, and this symbol of the relation should be something other than the symbols for felinity and animality. Thus if the symbol that expresses felinity were "cat" and the symbol that expresses animality were "animal," we should expect to find symbol strings such as *"the animal of the cat," *"the cat's animal" or *"the cat has an animal" reflecting animality's being part of felinity. We do not in fact find that such strings occur in that sense. Insofar as they occur at all, they signal ordinary possession; "the cat has an animal," fDr example, could easily be understood as referring to a situation in which a cat entered

Chapter Five

195

the room carrying a mouse in her mouth, but hardly as referring to the fact that being an animal is part of being a cat. Using the examples above, let us recapitulate the line of reasoning that Diimaga uses to show that words do not express universals or intensions. If "cat" and "animal" express the properties felinity and animality respectively, then "the animal, the cat" must express a wholepart relation between felinity and animality. But if a string of symbols expresses a whole-part relation, then that relation must be expressed by elements of the string other than the symbols for the relata in the wholepart relation. There is no symbol other than the symbols for the relata in the putative whole-part relation in the expression "the animal, cat." Therefore, "the animal, cat" does not express a whole-part relation between felinity and animality. Therefore, "cat" and "animal" do not express felinity and animality respectively. A line of reasoning similar to the above is used by Diimaga at PramalJasamuccaya 5:3 to show that a general term cannot be understood as expressing a relation between a universal and a particular in which it occurs. The view under consideration here is that a general term such as "cat" might be understood as expressing a relation between felinity and particular cats, specifically, the relation of felinity's occurring in particular cats. What makes this view plausible is that "cat" is not applicable to particulars in which there is no occurrence relation with felinity, and thus the applicability of the word is restricted to the occurrence relation, and being restricted to it, can therefore make it known to the hearer of the word. However plausible this account may seem, Dinnaga rules it out by appealing to essentially the same principle that he invoked to rule out the hypothesis that general terms express intensions or universals. Relations, when one intends to express them, are signaled by elements in a symbol string other than the symbols for the relata in the relation. In Sanskrit, for example, relations are expressed by sixth-form case-endings or special secondary suffixes (taddhita-pratyaya) that are attached to the symbol for one of the relata when one intends to express a relation and are not attached when one does not intend to express a relation. When, therefore, no such signal is found attached to a general term such as "cat," the general term cannot by itself be understood as expressing a relation, and when such a signal is attached to a general term, then it is this extra signal and not the general term that is expressing the relation.

196

Difmaga's nominalism

5.2.7 Particulars as instantiations of universals Dinnaga disposes of the above three hypotheses in short order, taking just two verses to eliminate the views that a word expresses its extension, that it expresses a universal and that it expresses a relation between a universal and all the particulars in which the universal occurs. It is noteworthy that in the second chapter of the Prama'J.lisamuccaya, in which Dinnaga discusses inferential evidence in general, he also argues that that about which a piece of evidence provides new knowledge is neither a universal nor a particular nor is it the relation between a universal and its particulars. Observing a particular instance of smoke, for example, serves to make known neither fire in general nor a particular fire nor the relation of fire in general to a particular locus, but rather it serves to make known a particular locus as qualified by the deniability of the absence of every particular fire. The arguments in that chapter, at PramalJaSamuccaya 2:8-11, are similar to but not identical with the arguments used in the fifth chapter to show that a general term does not express either a universal, a particular or the relation between the two. IS Here in the fifth chapter, however, a fourth hypothesis is discussed that, being somewhat more complex than any of the three considered so far, occupies a full six verses and their attending autocommentary. In these six verses Dinnaga argues against this hypothesis with three independent lines of argument. In discussing this hypothesis, I shall first present the hypothesis itself, then the three argl':ments against it, beginning with the most simple and ending with the most complex. DiIinaga himself presents the three arguments in a different order from the one I will use, but since the arguments are logically independent, the order is not important except perhaps for rhetorical considerations. The hypothesis under consideration states that a word expresses not a universal but an instantiation of the universal, that is, a universalpossessor or a locus of the universal in question. The main advantage of this hypothesis is supposed to be that it avoids the problem that was encountered above in connection with the unaccountability of predication when symbols that in isolation seem to express intensions occur in strings of symbols. According to this hypothesis that a word expresses a universal-possessor (fatimat vastu), a string in which predication occurs such as "the cat is an animal" or "the animal, cat" is to be understood as reading "the locus in which felinity occurs is a locus in which animality occurs or "x is a locus of felinity and x is a locus of animality." The word "cat" is capable of making a felinity-locus known, since "cat'~ is restricted in applicability to felinity-loci, and "animal" can make animality-loci

197

Chapter Five

known because it is similarly restricted in application. Moreover, a predication string such as "the animal, cat, since it is restricted to loci in which both felinity and animality occur, is capable of expressing the intersection of the set of felinity-loci with the set of animality-loci .. II.

In analysing the hypothesis that a symbol expresses a universallocus, Diimaga argues that there are only two ways to interpret the notion under consideration. First, the universal-locus may be regarded as a particular. But if this is how we regard it, then we face the problems mentioned above in connection with the hypothesis that a word expresses its extension, that is, each member of the set of particulars to which the word is applicable. 19 A second way to interpret the hypothesis under consideration is to take the hypothesis to be stating that a word expresses not a single particular universal-locus nor several universal-loci but rather the fact of universal-location, that is, the fact that is common to all loci of a given universal. But if this is our interpretation, then the hypothesis under consideration is saying in effect either that a word expresses a particular kind of relation, specifically an occurrence relation, or that a word expresses a universal.2° But we have already seen that there are problems in the view that general terms express intensions, and in the view that general terms express relations. Thus the hypothesis that a general term expresses a universal-locus is no advance over these other three hypotheses, all three of which Dirmaga has argued to be untenable. This is the simplest of the three arguments that Diimaga employs against instantiation as that which general terms express. We may now turn to a second argument that Diimaga uses against the workability of interpreting a general term such as "cat" as expressing that which possess the universal by virtue of which the members of the set of particulars to which "cat" is applicable are members of that set. If we adopt such an interpretation, says Dirinaga, we are saying in effect that the general term in question, such as "cat," is to be replaced by a qualifying phrase or adjectival construction, such as "felinity-possessing," whenever it occurs in a string. When this replacement occurs, each of the general terms in a given string of symbols comes to be replaced by a symbol that is adjectival in nature. A symbol can be considered adjectival in nature when it no longer functions as an independent symbol but rather is subordinate to some other explicit or implicit symbol that it qualifies. 21 Thus "the animal, cat" comes to be rewritten "animality-possessing, felinitypossessing [thing]," and in this rewritten string both "animalitypossessing" and "felinity-possessing" are functionally subordinate to a new head word, "thing," which mayor may not be expressed. Let us at this point introduce the two metasymbols "HN" and "ADJ" to represent a head

198

Diilllaga's nominalism

word slot in a symbol string and a subordinate word in a symbol string respectively. In the original phrase, "the animal, cat," the form of the symbol string is HN\HN2, whereas in the rewritten string, "animalitypossessing, felinity-possessing [thing]," the form is ADJ1ADJ2HNo. The principal question that now arises is whether the rewritten symbol string has preserved the information conveyed by the original string. If any information has been lost in rewriting the original string, then it must be conceded that we have not rewritten it properly. In answering this question we must fIrst determine what information the original string, HN 1HN2, conveys. According to Dinnaga, the structure of the original string clearly indicates a relation of a set to its subset, which relation he calls' "tadbhedatva." Membership in the larger set, indicated by one of the words standing in an HN slot in the above schema, automatically carries over (aksipati) to each member of the subset indicated by the word standing in the remaining HN slot. The structure of the schema HNIHN2 does not convey information as to which set is a subset of which. We may fInd either "the cat, an animal" in which the set picked out by the word in the HN \ slot is a subset of the set picked out by the HN2 slot, or "the animal, cat" in which the set picked out by the word in the HN2 slot is a subset of the set picked out by the HN] slot. But the principle in both cases is the same, that in an appositional phrase one of the head nouns will pick out a set and the other head noun will pick out a subset of that set. It seems to be Dinnaga's contention that the structure of a string of the form HN\HN2 clearly shows that the word in one of the HN slots is restricted in application to only but not necessarily all those things to which the word in the other HN slot is applicable. In the rewritten string, however, this relation of a set to one of its subsets is not clearly shown, for what we have in the rewritten string is just an implicit head noun qualifIed by two adjectives, the form of the string being represented by the schema ADJ1ADJ2HNo. In order to show that a string of this form does not necessarily show a set-subset relation between the two sets picked out by the words that fill the ADJ slots in this schema, it is sufficient simply to give an interpretation of the schema ADhADJ2HN in which the set-subset relation does not hold between the sets picked out by the two adjectives. It is not diffIcult to fInd such an interpretation. "The white, sweet sugar" is an interpretation cited by Diilllaga. Clearly "white" is not restricted to just those things to which "sweet" is applicable, nor vice versa. Membership in the set indicated by the word in the ADJ \ slot does not carry over to each member of the set indicated by the word in the ADJ2 slot nor vice versa. Hence there is not necessarily a set-subset relation between the adjectival symbols in the rewritten schema ADJ\ADJ2HN. And therefore this rewritten string does not preserve the information conveyed by,the original string HN]HN2. Because of this failure of the rewritten string to preserve

Chapter Five

199

the information of the original string, it is not legitimate to understand a general term such ~s "cat" to be synonymous with or expressing the same thing as or replaceable by an adjectival form such as "felinity-possessing." The rather elaborate argument outlined above would ~ot hold up, of course, if one could produce an interpretation of HN j HN2 that does not entail a set-subset relation between the set picked out by the word in the HN j slot and the set picked out by the word in the HN2 slot, for it seems to be presupposed by Diimaga that this appositional schema always indicates a set-subset relation. In fact, it is not much more difficult to produce a nominal apposition that represents an intersection of sets rather than a subset relation than it was to find an adjectival apposition that represented intersection rather than set inclusion. One example would be the phrase "the Indian, a logician,." which exemplifies two nouns in apposition and yet the extension of neither is a proper subset of the extension of the other, since some but not all Indians are logicians and some but not all logicians are Indians. This brings us to a third argument adduced by Diri.naga against the hypothesis that a general term expresses instantiation of a universal. The gist of this argument, which appears in Pramal)asamuccaya 5:4b-8b, is that if· a general term expresses instantiation at all, it does so only figuratively or indirectly, because the application of a general term to an instantiation can be only figurative or extended application. Once this is established, Diri.naga then sets out to show that the main principle of figurative application of words to objects is not at work in the case of general terms and instantiations of universals. From this lack of the workability of the principle of figurative application to the case under consideration, Diri.naga concludes that general terms do not apply to instantiations figuratively. But if general terms do not apply to instantiations even figuratively, then they do not express instantiations at all. With this outline of the overall strategy of Diri.naga's argument, we can now look at what he says in greater detail. It must be noted, however, that the presentation of this argument is very terse in the original, and there is no consensus on the part of the two Tibetan translators, Diri.naga's commentator Jinendrabuddhi, and the various critics of Diri.naga such as Uddyotakara and Kumanla Bhatt;a, as to how to interpret some of Diri.naga's statements in this section. The overall argument, however, is relatively clear and will be reproduced here in a form general enough to avoid the particular problems involved in Diri.naga's presentation.22 The first point that Diri.naga makes clear is that if a general term applies to an instantiation at all, then this application can only be figurative

200

DiIinaga's nominalism

or secondary. Backing up this claim is the observation that some notion of a universal is logically antecedent to or is presupposed by the notion of a possessor of that universal, for there can be a universal-possessor only if there is a universal to be possessed, but it is not the case that there can be a universal only if there is a possessor of that universal. Thus, before the general term "cat" can generate the notion of felinity-possessor, it must first generate the notion of felinity itself, and hence felinity must be regarded as the primary or literal object of expression of the word "cat," and any reference to that which possesses felinity or that in which felinity occurs must be secondary. This observation would presumably be consistent with the views of the grammatical tradition at DiIinaga's time. In fact, Jinendrabuddhi reports that among some grammarians the application of a word to an instantiation of the universal named by that word would be even more removed than secondary application. According to this view, the primary object of a symbol token such as "cat" is the symbol class "CAT". "CAT" expresses felinity, which is then a secondary object of expression of "cat". The felinity-possessor is then a secondary object of expression of "CAT" and a tertiary object of expression of any given token of that symbol class. Thus, taking this view into account, we can say that a general term's applicability to an instantiation is, according to the linguistic philosophers of DiIinaga's time, at best secondary. The second point that DiIinaga makes is that the principle by which a given symbol can be said to apply figuratively to a given object is not seen to be at work in the case under consideration, that is, where a general term is allegedly figuratively expressing a universal-possessor. The principle in question here is that a symbol S that applies literally to an object L can apply figuratively to a secondary object F only if F is qualitatively similar to L. For example, we are justified in applying the word "cat" in some figurative sense to, say, a thief only if cats and thieves have some qualitative similarity, such as the ability to walk very quietly and to work skillfully in the dark. This qualitative similarity between the literal and the figurative objects of reference of a symbol need not even be a real similarity, that is, a similarity based on a quality's actually occurring in each of the two objects serving as the literal and the figurative objects to which the symbol is applied. Even if the quality on the basis of which a symbol is applied figuratively only appears to belong to the object of the figurative application, we can still apply that symbol figuratively. We can, for example, apply the term "blue" to a mirror that only appears to have the quality of blue colour because it is placed near an object that actually has the quality of blue colour. But whether the qualitative similarity between the literal object and the figurative object be real or apparentpit must be present in order that we may figuratively apply a word that literally applies

Chapter Five

201

to one object to another object. Hence, if we are to apply a general term, which applies literal.Iy to a universal, to a universal-possessor, we must find some real or apparent similarity between the universal and the universal-possessor. Now if there were such a similarity, aIgues Diimiiga, we should be able to isolate it and name it independently, and we should be able to apply this name for this shared quality in turn to each of the things that possess it. We can, for example, readily find what quality a conch shell, a jasmine flower and a lily have in common, namely, the colour. white. And we can apply the name for this quality to each of the things that possess it. We can say "the white conch" and "the white jasmine flower" and "the white lily" and so forth. Similarly, if there is a qualitative similarity that a universal such as felinity has with the universal-possessor such as the felinity-possessing cat, we should be able to name it readily. But according to Dmniiga, there is no such readily nameable similarity, or at least not a real similarity, on the basis of which we can apply the felinity-expressing word "cat" to a felinity-possessing cat. 23 Nor, of course, can we find a basis for predicating a term of more general scope, such as the animality-expressing word "animal," to a cat. Suppose it is argued, however, that the instantiation of the universal, by virtue of the fact that the universal is present in it, appears to possess certain features of the universal itself, and that it is on the basis of these features that thus appear to belong to both the instantiation and the universal that we figuratively apply the word for the universal to the instantiation. The instantiation would be analogous to a mirror that reflects the visible properties of what. is placed near it and by virtue of these reflected properties appears to have a qualitative similarity to the object in which these qualities actually occur. If the above analogy is to be taken seriously, we are still faced with the problem of determining which of the many characteristics of the universal appear to occur in the instantiation to enable us to apply the word that literally expresses a universal figuratively to the instantiation. There are essentially two distinct solutions to this problem that might be advanced. We may either say that all the characteristics of the universal are reflected in the universal's instantiation, or we may say that only some of them are reflected. The first of these two solutions is held to be untenable on two grounds. First, if all the characteristics of the universal appeared to occur in its instantiations, then it would be possible to have distinct cognition of an instantiation qua instantiation independently of a distinct cognition of the universal alone. It would be possible, for example, to know of a cat that it is a cat without first having to know what felinity is, for all the characteristics of felinity would appear to occur in the cat just as all the

202

Diilnaga's nominalism

visible qualities of a reflected object appear to occur in the mirror that reflects it, since reflection enables us to know the visual qualities of the reflected object without knowing the reflected object independently. It is clearly absurd, however, to say that we know of a cat that it is a cat without first knowing what felinity or cathood is. A second grounds for denying that all the characteristics of the universal are reflected in its instantiations is that if all the characteristics of the universal appeared to be shared by its instantiations, then the universal and the instantiations would have, or appear to have, exactly the same set of characteristics, including the characteristics of being at the same place at the same time. But if two entities A and B have exactly the same characteristics including spatial and temporal location, then there is no longer any justification for claiming that they are two distinct entities, and we have no grounds on which to deny that a universal is identical to its instantiation--not only qualitatively identical but numerically identical. So, if we wish to hold to the claim that universals are distinct from particulars, we must concede that not all the characteristics of universals are or appear to be shared by particulars in which universals OCCUr. 24 But if only some of the characteristics of a universal are reflected in its instantiations, we must still ask which of those characteristics are so reflected. Any attempt to answer this latter question presupposes that we can isolate the characteristics of the universal from one another. This presupposition Diilnaga denies. Objects appear in our awareness, he argues, as unified wholes, not as isolated characteristics. When we see a black kettle, for example, we see a black kettle all at once; we do not see its black colour, its shape, its kettlehood, its substantiality and so on as isolated characteristics in a sequence of cognitions. Similarly, when we are aware of a universal, we are aware of all its features at once, and because of this simultaneity of our awareness of all its many features, we cannot isolate some of those features and say that just those are transferred to or reflected in the instantiation. Thus we must ultimately reject the argument that a universal's characteristics are somehow reflected in its instantiations by virtue of the proximity of the universal and the instantiation. DiIi.naga's conclusion, then, is that it makes no sense to say either that all or that only some of the characteristics of a universal either actually or only apparently belong to its instantiation. And without some common feature between the universal and its instantiations, we have no ground for applying a word for a universal figuratively to its instantiations. But if a general term does not apply to instantiations even figuratively, then it does not express them at all. With these three lines of argument against the contention that general terms may be understood as expressing loci as instantiations of universals, Diilnaga closes off his discussion of the

Chapter Five

203

four principal views that rival his own view, which is that general terms express apoha. Up to this point we have concentrated our attention on just one type of symbol, general terms, despite the fact that Diimaga himself mentions five types of word. 25 The five types of symbol that he recognizes are general terms (jiltisabda), adjectives (gw;asabda), verbs and deverbatives (kriyilSabda), terms for relata in specified relationships (cf. dravyasabda), and proper names or singular terms (yadrcchilSabda). The first four of these five types of terms clearly pertain to repeatable characteristics such as properties, generalities and relations. But according to Diimaga even the fifth category pertains to generalities in the sense that an individual is regarded as a mental synthesis of a multiplicity of particular cognitions. Thus, as B.K. Matilal has pointed out, all terms are ultimately general terms in the sense that they are all applicable to a multiplicity of cognitive images. 26 The only distinction between the so-called general terms and socalled singular terms is that the latter have a narrower range of applicability than the former. But in this, the relationship between general terms and singular terms is in principle no different from that between any general terms of different scope, such as for example "animal," "dog," and "bloodhound." For this reason, Dirinaga takes general terms (jiltisabda) as paradigmatic for all symbols from which information can be derived, and thus he deals almost exclusively with this type of term throughout the discussion of symbols in the chapter on the apoha theory of meaning. To recapitulate what has been said up to this point of the overall argument of the apoha chapter, it was first argued that a linguistic sign (sabda) serves as an inferential sign (lin-ga) to produce in the hearer of the symbol knowledge that the subject of discourse, the thing to which the speaker of the symbol is applying the symbol, has a given property. This knowledge in the hearer arises through a cognitive process that can be described as an inference of the form: (2) If symbol S is restricted in application to property R, then if symbol S applies to locus x, then property·R occurs at locus x.

The next line of inquiry amounted to an investigation into the nature of our knowledge when we know that a given property occurs at a given locus, and the central question was: what is the content of the knowledge that the hearer of a symbol gets about that to which the symbol is applied? In answering this question it was argued that when we say that a general term expresses or makes known a property, then we are ultimately compelled to adopt an apoha interpretation of properties, because various facts of

204

Dinnaga's nominalism

language and principles of epistemology preclude all other ,plausible interpretations of what it is that a word expresses. Let us now use a concrete example to show what rs involved in adopting an apoha interpretation of properties. Let us use as our general term the word "human being" and use an individual whose name is Devadatta as an object to which the general term "human being" is applicable. According to what has been said of the form of the inference in which a linguistic sign is used as an indicator of some property, one infers as follows: "If the symbol 'human being' is restricted in application to humanity, then, if 'human being' applies to Devadatta, then humanity occurs at Devadatta." Now, eliminating the reference to the property mentioned in the conclusion of that inference by replacing the name for that property with the appropriate apoha translation, the conclusion of the inference becomes: "It is not the case that no instance of humanity is at Devadatta." Next, we can take an expression of the form "x-hood is aty" to be saying in metaphysical terminology what is said in more ordinary discourse by an expression of the form "x is y" or "y is x." Accordingly, we can express the above conclusion as follows: "It is not the case that Devadatta is no human being." This statement is taken by the apoha theorist to be the most accurate representation of the form of the information learned from hearing the sentence "Devadatta is a human being." 5.2.8 Absurdities in the view that universals exist outside thought

In saying that the sentence "Devadatta is a human being" conveys the information that it is not the case that Devadatta is no human being, the point is made that the property that the word "human being" in that sentence expresses is not the universal humanity but rather the deniability of the absence of each particular instance of humanity. But of course in stating things in this way we have not eliminated the need to have recourse at some point in the explanation to the notion of the universal humanity. But if, as Dinnaga argues, there are absurdities in the hypothesis that a universal is a single entity that resides in each of many individuals within its extension, then it becomes necessary at some point to propose a plausible alternative to this hypothesis. That the truth of the sentence "Devadatta is a human being" consists in its being the case that Devadatta is a human being is a matter over which there is not likely to be much controversy, but what has to be settled is the question, in what does the fact of Devadatta's being a human being consist? If one is committed to denying that the fact of the world underlying the truth of the sentence

Chapter Five

205

"Devadatta is a human being" is the residence of the entity humanity in the entity Devadatta, then it is incumbent upon one to offer a different account of what fact of the world it is that underlies the truth of that sentence. As we shall see presently, the conclusion towards which Diimaga argues is that the fact underlying the truth of the sentence "Devadatta is a human being" is no more than the fact that, by whatever conventions there may be that govern the use of the symbols in a given language, the equivalent in that language to the word "human being" is applicable to the individual Devadatta. Devadatta's so-called humanity consists only in the fact that those who know how to use, for example, the word "human being" correctly in English would agree that the individual Devadatta is among the objects of the universe °to which it would be appropriate to apply the . word "human being." Now if this account of truth that Diimaga advocates is carried through, it turns out that the sentence "Devadatta is a human being," which was said above to express its not being the case that Devadatta is no human being, is now taken as expressing only its not being the case that "human being" is inapplicable to that to which it is not the case that "Devadatta" is inapplicable.

5.2.9 The contrariety of expressions The context in which Diimaga argues for the nominalist position we have just outlined is in his treatment of contrary terms. This discussion begins at Pramii1J.asamuccaya 5:25cd, where it is observed that if symbol S is applicable to locus x, then we may know that locus x has the property of not being a candidate for the applicability of a symbol T such that T is other (anya) than S. This naturally raises the question of what it means for one symbol to be "other" than a second symbol. Clearly we do not want to say that if a given symbol is applicable to an object then no symbol that is of a different symbol class is also applicable to that object, for this would prevent us from making any predications at all. We could not, for example, say "The cat is meowing," because "cat" and "meowing" are not tokens of the same symbol class, and thus could not both apply at the same time to the same object. Thus Dirinaga says in Pramiil].asamuccaya 5:25cd that a term's applicability to a given object does not rule out the applicability of 1. terms that are synonymous with the term in question 2. terms that are wider in extension than the term in question 3. terms that are narrower in extension than the term in question. In this context a term A is wider than a term B in case A is applicable to every object to which B is applicable but there are things to which A is

206

Dirinaga's nominalism

applicable to which B is not applicable, or in other words if B's extension is a proper subset of A's extension. Given all this, it is clear what Dirinaga is. saying is that if a given symbol applies to a given locus, then some linguistic signs that are of a different symbol class are not also applicable to that locus. The obvious question that now arises is: of all the symbols that are different in form from the symbol that is applicable to a given object, which ones are not applicable to that object? In short, how do we determine that the two terms are each other's contrary terms? In considering the above question, Dirinaga offers two different hypotheses to account for the contrariety of terms. The first hypothesis that he considers, and ultimately rejects, is that the contrariety of terms has its basis in certain metaphysical facts or certain natural features of the world that are ultimately independent of our discourse about the world. The second hypothesis that he considers, and ultimately accepts, is that the contrariety of terms is no more than an essentially accidental feature of linguistic convention from which we can draw no firm conclusions about the nature of the world. The first of the two hypotheses mentioned above is outlined in Pramar;asamuccaya 5:28-30. Underlying it is the supposition that the different classes as recognized by the Sanskrit grammarians reflect distinct and real metaphysical categories. The Sanskrit grammarians divided words into four classes, claiming that each class had a different grounds of application (pravrttinimitta) to an object. According to this view, the grounds for applying a general term to an object is that object's possessing a universal (jati). The grounds for applying an adjectival term to an object is that object's possessing a quality (gur;a). The grounds for applying a verbal term to an object is that object's being the locus of an action (kriya). And the grounds for applying a singular term to an object is that object's individuality. In their classification of words and their respective grounds of application, the grammarians take no stand on the question of whether it is reality that determines how we classify words or our classification of words that determines how we classify our experiences of the world.

There were philosophers, however, who were decidedly committed to the view that it is reality and its natural divisions that determine our classification of terms. Among these realists were the Naiyayikas and Vaise;;ikas, whose views Dirinaga most probably had in mind when outlining the hypotheses now under discussion. According to the Nyaya and Vaise~ika schools, the real things of the world fall naturally into mutually exclusive ontological categories, the most important of which for the present discussion are substance (dravya), quality (gwJa) and action (karman). The substances themselves fall naturally into various

Chapter Five

207

subcategories in accordance with the general and specific properties inhering in them and in accordance with the primary elements of which the substances are composed. Thus the universe of substances has, in this view, four mutually exclusive classes, namely, the class of earthen substances, the class of watery substances, the class of fiery substances, and the class of airy substances. Each of these four subcategories is again divided into natural and mutually exclusive genera, and each genus is divided into natural and mutually exclusive species, which in tum are divisible into subspecies and so on. According to the so-called realists of Indian philosophy, such as the Naiyayikas and Vaise~ikas, it is this natural hierarchy of genera and species that accounts for the fact that some terms are contrary to others. The reason that wider terms are not contrary to their narrower terms is that the wider term names a general property that defines a class that wholly contains the species defined by the specific property named by the narrower term. Two terms that are both narrower than a given wider term, however, are contrary to one another, because the genus named by the wider term is divided into mutually exclusive species; since no object that belongs to one of these species has the specific property that defines the other species, no term naming the specific property that defines one of the species is applicable to any member of the second species. Thus any term of which the grounds of application (pravrttinimitta) is a property that defines a given class is directly contrary to that other term of which the grounds of application is a property that defines a second class that is disjoint from that given class. Sometimes, however, the contrariety of terms is indirect rather than immediate. Indirect contrariety occurs when a term of which the grounds of application is a property that defines a species contained in one genus is contrary to a term of which the grounds of application is a property that defines a species contained in another genus. Thus, for example, the term "0a...1c" may be seen as indirectly contrary to the term "porcelain," because "oak" names a species of the genus tree, which is mutually exclusive with the genus pottery of which porcelain is a species. The notion of .indirect contrariety just outlined raises some problems, one of which Dir'maga discusses. If, as the hypothesis maintains, words that name specific properties of two different categories are indirectly contrary, then We should expect to find that a term such as "oak," which names a property that defines a species of a genus in the category of substances, is indirectly contrary to a term such as "green," which names a property that defines a species in the genus colour which is in tM category of qualities; for qualities and substances are mutually exclusive categories of the supercategory reality (sattva). Thus we should not be able to say "The oak is green" without uttering a contradiction

208

Dinnaga's nominalism

arising from applying contrary terms to one and the same object. Suppose that, in order to avert this difficulty, it is said that there is a limit to how far indirect contrariety extends. Suppose we say that a specific term is indirectly contrary only to the genera that are mutually exclusive to its own genus but not to the species of those genera. Saying something like this does avert the problem it is supposed to avert, but in doing so it allows a number of expressions that we intuitively feel should not be allowed. There would, for example, be no harm in saying "The oak is a colour," since "colour" names a species of quality and would not be indirectly contrary to "oak," the indirect contrary of which extends by this principle of limited contrariety only as far as the term "quality" itself, quality being the genus directly contrary to the genus substance to which oak belongs. Perhaps one could, by introducing further modifications into the principle of contrariety, arrive at some statement of the principle that would account for all and only those sets of terms that we intuitively feel to be contraries. If one could arrive at such a principle, then one could claim that the principle so derived is true in virtue of its reflection of extra-linguistic realities, that is, of the world itself as opposed to merely our way of talking about the world. But such a claim would be difficult to maintain in the light of the fact that our guidelines in arriving at the principle would not be our observations of the world itself but rather our intuitions--or even our systematic observations--about the usage of words. The givens on the basis of which we arrive at a theory of which terms are contrary to which would be facts about the conventional usage of those terms, and the assertion that conventional usage of terms reflects the structure of extra-linguistic realities remains itself an unproven claim. It is for this reason that DiIinaga ultimately rejects the hypothesis that the contrariety of terms is founded upon metaphysical realities. How then does Dinnaga propose to account for how it is that we know that some terms are contrary to others? The answer, already anticipated in the discussion above, is spelled out in Prama/')asamuccaya 5:31, where it is indicated that we know that one term is contrary to another by virtue of our observations of the usage of the two terms in question. If one term is never observed to be applied to objects to which another term is applied, then the terms are contraries. There is no need to ask why it is that, for example, the word "oak" is not applied to any Qbject to which the word "vase" is applicable, for it is sufficient simply to know that the two terms have disjoint extensions. If pressed to answer why the two terms have disjoint extensions, we can only say that this is due to the fact that our linguistic conventions are such that the two words are not conventionally applicable to the same range of objects. But there is not necessarily a basis in reality for our conventions being as they are.

Chapter Five

209

Our linguistic conventions may tum out to be quite arbitrary, or at least to have no justification of which we are aware. Dirinaga illustrates this latter point in Pramtit;lasamuccaya 5:37d. Here he imagines an attempt to account for why it is that "blue" and "yellow" are contrary to each other but not contrary to "colour," whereas "blue" and "sweet smelling" are not contrary to each other· despite the fact that their so-called genera, colour and odour, are contrary to each other. The attempt to account for such things is, he argues, doomed to failure. What property do objects to which "blue" is applicable have in common with objects to which "yellow" is applicable? We might say that they have in common the fact of being some colour or of having some colour. What do objects to which we apply the word "square" have in common with the objects to which we apply the word "round"? We might say that they have in common the fact Df being or having some shape. What do objects to which we apply the word "colour" have in common with objects to which we apply the word "shape"? We might say that they all have in common the fact of possessing visible qualities (rupa). We might then conclude that blue, which is the property that all objects to which we apply the word "blue" have in cornmon, has in cornmon with yellow, which is the property that all objects to which we apply the word "yellow" have in commDn, the fact of being a visible quality. And since this fact of being a visible quality is absent in the property on the grounds of which we apply "sweet smelling" to an object, we can safely conclude that colours and odours are different genera. But suppose we ask what it means to say that something is a visible quality. It might be answered that a visible quality is a property that enters our cognition by means of the eye. Does this mean, then, that every property that enters our cognition by means of the eye is a visible quality? If so, then surely the substance in which the visible quality such as blue is said to inhere is also a visible quality, for it is surely as visible as the blue colour that inheres in it. Why then do the metaphysicians insist that blue colour is a visible quality, while the flower in which it supposedly resides is not a quality at all but a substance? It is Dirinaga's contention that, whatever the metaphysician might say to justify his division of the world into the categories described in his system, the real reason is that he is being guided along by his observations of linguistic usage and nothing more. That is, he puts a flower in the category of substances and the colour blue into the category of qualities simply because by convention we apply the word "substance" to every object to which we apply the word "flower" and we apply the word "quality" to every object to which we apply the term "blue colour." Similarly, he regards blue as belonging to the species of qualities that are visible but sweet smell as belonging to the species of

210

Diimaga's nominalism

qualities that are aromatic simply because objects to which ,"blue" is applicable are objects to which "visible" is applicable and objects to which "sweet smelling" is applicable are objects to which "aromatic" is applicable. It is Diilnaga's contention that it is not the case that a group of objects fall together into a class on the basis of their all having a certain property or properties in common and that one expresses that class by means of a term that names that common propert;'; but rather, a group of objects fall together into a class solely in virtue of the fact that one applies the same term to each of the objects in the group. A word does not name a universal that defines a class. Rather, the word itself defines that class. The word itself is the universal. One possible objection that Diilnaga's contemporaries might be expected to raise against his nominalist position is that classes formed on the basis of natural universals are bound to be far more stable than classes formed on the basis of arbitrary manmade conventions. The membership of a particular in a class is a straightforward and predictable matter when that membership is determined by the observable or inferable presence or absence of a universal in the particular, and this predictability of particulars is a prerequisite to understanding the world and acting rationally towards it. To make membership in a class depend on something as relatively capricious and volatile as human convention is to render classes far less stable and to render concepts that have classes as their contents less suitable to the task of guiding human behaviour in a world of particulars. And since the only world in which behaviour is possible is in the concrete world of particulars, to be without principles to guide one through that world is to be without something that would be very valuable indeed to have. Anticipating an objection of this kind, Diilnaga sets out at PramiifJasamucccrya 5:36 to discuss the attributes that universals are supposed to have that enable them to serve as the basis of stable classes and class concepts. Three attributes had been assigned by the Indian tradition to universals: a) unity, b) constancy and c) residence in a plurality of particular loci. These three attributes of a universal taken together were seen to enable the universal to function as a simple principle in accordance with which a number of particulars can be treated together in a single thought or referred to by a single symbol, a principle that does not vary from one particular to another and does not undergo changes as its particulars do. As has been mentioned earlier, Diilnaga argues that if a universal be regarded as a real entity that physically resides in its particulars, then it cannot be both a single entity and simultaneously resident in a plurality of substrata. There cannot be an entity that has all three of the attributes of a universal as described above. If, however, we abandon the notion that a universal physically resides in particulars, saying

Chapter Five

211

instead that it is simply a conceptual construct that the intellect superimposes upon particulars, then the above three attributes--or rather, suitably modified forms of them--cease to be incompatible, for there is no absurdity involved in saying that a single idea can be about many things. The modified notion of a universal, then, becomes that of a single conceptual abstraction that the mind can apply to a plurality of particulars, a conceptual abstraction that persists despite changes in state that may occur in the particulars upon which it is superimposed. But if it was possession of the three attributes of unity, constancy and plural residency that enabled universals to serve as the basis of classes, then possession of three very similar attributes by a conceptual abstraction should equally enable that abstraction to perform the same function that universals were supposed to do. Hence, concludes Dirinaga, a linguistic and epistemological theory is not worsened by replacing universals with conceptual constructs. On the contrary, it is improved for the reason that the belief in conceptual constructs, unlike the belief in extraconceptual realities, entails no absurdities. As we have already seen, the abstraction that the intellect superimposes upon particulars is, according to Dirinaga, the notion of class membership determined by the applicability of a given linguistic sign. A given symbol S, in being applied correctly to an object by an accepted convention, expresses the fact that a) the object is a member of the class of objects to which S is applicable, b) the object is not a member of the class complementary to the class of objects to which S is applicable, and c) the object is not a member of the class of objects to which any symbol contrary to S is applicable. Throughout the fifth chapter of the PramiiJ)asamuccaya the point is made that a verbal symbol (sabda, sgra) serves as an inferential sign (liliga) to produce in the hearer knowledge that the subject of discourse, the thing to which the speaker of the symbol is applying the symbol, has the property of being within the extension of some symbols and outside the extension of other symbols. The basic claim behind the apoha theory of meaning is simply that every symbol divides the universe into two and only two mutually exclusive classes, and from this it follows that a symbol marks the segregation (apoha) of the members of its own domain from the complementary class and that it marks the exclusion (apoha) of all contrary symbols from its own domain. This basic claim is a very fundamental one in two senses. First, it is fundamental in the sense that no one, regardless of her metaphysical positions or her commitments to various theories of-language, is very likely to dispute it. Where controversy does arise, as we have seen, is over the question of whether we are entitled to claim that a symbol expresses more than this very fundamental principle

212

Diimaga's nominalism

of radical segregation. Dinnaga's position has consistently been that we are not so entitled, and it was this position rather than the basic claim behind the apoha theory that brought criticisms from so many sides. A second sense in which this basic claim is very fundamental is that the central observation in the apoha account Df a symbol's function and its relation to other symbols and to that which is symbolized holds true regardless of what type of symbol may be under discussion. Thus this schema If S is applicable to x, then if S and T are contrary symbols, then T is not applicable to x.

remains good whether we interpret "S" and "T" as terms or as whole propositions, or indeed as whole texts. Moreover, if we interpret "S" and "T" as terms, then the schema remains good whether we interpret them as adjectives, verbs, general terms, proper names, syncategorematic terms, case endings or whatever other grammatical or logical categories of symbols we may wish to use.

S.2.10 The meaning of a sentence As was indicated above, despite the fact that the basic claim behind the apoha theory contains within it no predisposition for one or another metaphysical theory, Dinnaga himself argues for a nominalistic ontology and incorporates these arguments into his presentation of the apoha theory. Similarly, although the basic claim behind the apoha theory contains no predisposition to regard one set of symbols as primary and all other sets as secondary, Dinnaga himself goes on to argue that one kind of symbol is more basic than any other. The primary symbol, he claims at Pramar;asamuccaya 5:47-49, is the sentence. All other sorts of symbols, such as proper names and general terms and verbal predicates and case endings and syncategorematic terms, are secondary in the sense that they are derived from the sentence. Diimaga deals with two questions in connection with his assertion that the basic symbol, the primary unit from which all others are derived, is the whole sentence. First, he deals with the question of grounds for making the claim at all. And second, he deals with the question of what the basic unit of meaning is that is made known through this basic symboL Since part of his answer to the first of these questions depends on his answer to the second of these questions, we shall take up the second question fIrst. The basic unit of meaning that a sentence makes known to its audience is called by Dmnaga, who follows Bha:rtrhari's terminology, pratibha. The word "pratibha" literally means a reflection or appearance

Chapter Five

213

or a vivid image, but as Diri.naga uses it this term signifies the idea of cognitive content of a single cognition, the comprehension in a single act of awareness of a complete state of affairs. Although a complete state of affairs may be broken down for the sake of analytic understanding into constituent parts, it is initially grasped spontaneously and immediately in a single cognitive act of which the content, the pratibhii, like the single cognitive act itself, is a single whole undivided into internal divisions of any kind. In connection with this view, there are two questions to which Diri.naga addresses himself in the fifth chapter of the Pramcll;asamuccaya. First, there is the question of why the pratibhii, which being the primary content of awareness serves as the most fundamental datum upon which all further thought is based, must be regarded as a simple unity and not as a complex whole made up of parts. And second, there is the question of the relation between pratibhii as the object made known through a symbol and apoha as the object made known through a symbol. The principal argument behind the claim that a state of affairs that serves as the content of a cognitive act must be regarded as a simple, nonpartite object rather than as a complex comprising a multiplicity of aspects is to be found at PramiilJasamuccaya 5:17-19 and the autocommentary thereto, a passage that we have had occasion to discuss in previous contexts. In its most general form, the conclusion of the argument contained in this passage is that the relation between a single individual and a multiplicity of attributes cannot be a real relation, because there are only two possible ways of characterizing the relation, neither of which is compatible with the claim that there is a real relation between a single individual and a multiplicity of attributes. The two ways of characterizing the relation are: a) to say that every attribute occurs throughout the individual, that is, in every physical part of the il).dividual, or b) to say that each attribute occurs in its own particular substratum, each substratum of an attribute being a physical component of the individual that putatively has a plurality of attributes. If the first of these two characterizations is accepted, then there can be no truth to the claim that the attributes are distinct from one another; each attribute, having exactly the same physical and temporal location as every other, becomes indistinguishable from any other attribute. But if the second of the above characterizations is accepted, then there can be no truth to the claim that a plurality of attributes are occurring in a single substratum, namely, the individual. Each attribute having its own private substratum, the putative individual to which the plurality of attributes supposedly belongs loses its integrity. Therefore, says Diri.naga, it is not the case that a) a plurality of attributes can reside in a single substratum,

214

Dirinaga's nominalism

nor that b) a single attribute can reside in a plurality of substrata. As we have seen in previous discussions, the denial that a single attribute can reside in a plurality of substrata lies behind the denial of universals as realities that exist outside the mind. In the present discussion, the denial that a plurality of attributes can reside in a single individual lies behind the denial of cognitive images having a plurality of aspects. What this means is that if one has a single cognition of some state of affairs such as that one red book stands to the left of two green books on the surface of a black desktop, then this entire state of affairs is to be taken as a single datum in cognition, one single pratibhii. A speaker, wishing to convey information concerning the contents of her awareness of a state of affairs, can do so by employing a symbol in the form of a sentence. For a given state of affairs, such as that of one red book standing to the left of two green books on the surface of a black desktop, there are normally a variety of symbols that may be used. For the state of affairs just mentioned, here are three possible symbols from among ~any:

1. "A black desktop is under three books, of which two are green and one is red, the red standing to the left of the green." 2. "Two green books are to the right of one red book standing upon the black top of a desk. " 3. "The single book standing to the left of two green books on the black desktop is red." Each of these symbols divides the universe into two mutually exclusive classes. In this context, the universe is the set of all states of affairs, and it is divided into the set of those states of affairs to which the sentence in question applies and the complementary set, which is the set of all states of affairs to which the sentence in question is not applicable. The hearer of one of these symbols knows upon hearing it that the subject of discourse, the state of affairs that is the content of the speaker's awareness, is a member of the set of states of affairs that satisfy the sentence spoken. An object is said to satisfy a symbol iff the symbol is applicable to (is true of) the object. But, since the sentence does no more than exclude some states of affairs of the universe from the domain of present discourse, any number of hearers may form any number of mental images of their own upon hearing the sentence. But do all·these mental images have anything in common? Can there be anything in common between the idea in the mind of the speaker of a sentence and the ideas formed by the sentence in the

Chapter Five

215

minds of those who hear it? According to the apoha theory of meaning there is a limited .amount of commonality. What these private mental images generated by the sentence in the minds of the hearers will have in common is simply that all satisfy the sentence spoken, or to put it another way, none satisfies any sentence that is logically contrary to the spoken sentence. Thus, says Diilnaga, in one sense the meaning of a sentence is the line of demarcation between the states of affairs that satisfy it and the remainder of the universe. This is the meaning of the sentence considered abstractly. In another sense, from the personal point of view of anyone hearer of the sentence, the sentence's meaning is the state of affairs that that listener is able to picture in his mind upon hearing the sentence. 5.2.11 The sentence as the primary linguistic symbol Let us now tum to the question of the grounds for Diilnaga's claim that the whole sentence rather than the individual word is the fundamental unit of speech from which others are derived. Both arguments are found in Diilnaga's own prose commentary to Pram[u;asamuccaya 5:46. The first argument is to the effect that an individual word has no meaning except in the context of a sentence. The reason for making this claim is that words have no use except in sentences. A single word as an item in a lexicon does not serve to communicate any information, but as soon as it is employed in a sentence, then the hearer of that sentence begins to form ideas about what the speaker of the sentence is thinking. This is not to say that one cannot establish some sort of meaning for an individual word by analysing how it is used in a variety of contexts, but the claim is that one cannot even begin this process--not to speak of arriving at any successful results--until one understands the sentences in which that word is used. The meaning of a word, then, is derived from the meanings of the sentences in which it occurs and must be regarded as something of a fiction, albeit a fiction that has a variety of useful applications. In this respect a word is analogous to any other fragment of sound that may occur within an entire sentence; such as a prefix or suffix. Although it makes no sense to ask what the sound lsi means by itself in English, one could isolate a number of ways it occurs in sentences, and by -analysing how the meaning of the sentence "The vat leaks" differs from the meaning of "The vats leak" and "The vat's leak (needs repairing)" and so forth, one could derive a number of meanings for this single phoneme. The derivation of the meaning of other fragments of sentences, those fragments such as "vat" and "leak" that we call words, is in principle no different from the derivation of meaning for those fragments called case endings or inflectional affixes. And so, argues Diilnaga, since words are like

216

Diimaga's nominalism

inflectional affixes in that they serve no function except in the context ofa sentence, and since knowledge of the function they do serve is dependent on a knowledge of the meanings of the sentences in which they occur, words are not to be regarded as the fundamental unit of speech from which more complex units are made up. Rather, sentences are to be regarded as the basic units of meaning from which the meanings of its audible components are derived by intellectual exercise. The second argument that Dirinaga offers in support of this claim is that a sentence is the symbol for the pratibhii, the basic unit of meaning, and the pratibhii is nqn-partite, and therefore the symbol that represents it must be non-partite. Internal divisions may be superimposed by the faculty of reason upon a complete state of affairs, and similarly internal divisions may be superimposed by the faculty of reason upon the complete symbol that is the sentence. But in both cases, it is the whole that is the fundamental datum, the material out of which the intellect fashions its understanding of the world. It is only when that whole, whether it be the whole state of affairs or the whole sentence that represents it, is not understood that the faculty of reason breaks it into parts and studies the relations between those parts. This process of analysis is a strategy that one may employ either to deepen one's own understanding of the whole or to explain it to others.

-- Notes--

Chapter 5. Dilinliga's nominalism

1. Abhidharmakosa 6.4. In Vasubandhu 1972 ed., p. 889. yatra bhinne na tadbuddhir anyapohe dhiya ca tat ghatlimbuvat samvrtisat paramarthasad anyatha 2. Vasubandhu 1972 ed., p. 80: "kal) punar vi~ayalambanayor vise~aQ. yasmin yasya kiiritrarl! sa tasya vi~ayaQ. yac cittacaittair grhyate tad alambanam." 3. What I have offered here is a paraphrase of the information contained in Yasomitra's commentary to the AbhidharmakosabhiLua in Vasubandhu 1970 ed., p. 80. The passage from which this information has been gleaned is:' "yasmin yasya karitram sa tasya vi,aya iti. kiiritram puru~araI). c~uQsrotrii.dInam riipaSabdii.di~v alocanaSravaI).adi kiiritram.

Chapter Five

217

tac ca svacittacaittan pratyasrayabhavaSaktivise~alak~aJ)am veditavyam. yac cittacaittair grhyate daI).oavastambhanayogena tad iilambanam riipadi. tad evam sati cittacaittanam evalambanam. vi~ayal:t punaS cak~uradlnam api, na kevalam cittacaittanam. n 4. The Sanskrit original of the Alambanaparfk'fii does not survive, but of relatively classical versions of the text there are one Tibetan and three Chinese translations. My translation here is based upon the text of the Tibetan translation of the Alambanaparfk'!iivrtti as edited by Tola and Dragonetti 1982:120-123. Another edition of this anonymous Tibetan translation appears in Frauwallner 1959b:157-161. Tola and Dragonetti's study of the text, like Frauwallner's, records all the Sanskrit fragments that have been found cited in still extant Sanskrit works. Their study also includes an annotated English translation of the verses with Diflllaga's own prose commentary. Their translation is rather cumbersome, but their study of the next is informative, although flawed by an outdated interpretation of Diflllaga's position as one of subjective idealism. A somewhat smoother English translation appears in Sastri 1942, which also has a translation of extracts from the commentary by Vinltadeva. There is also an English translation of just the verse text of Alambanaparfk'fii in Wayman 1979, and although his translation itself is relatively poor, Wayman's depiction of Diflllaga's overall philosophical stance is on the whole better informed than depictions found in other English works on the subject. To my mind the clearest rendering of the text is the translation into French by Yamaguchi 1929, which also benefits by references to the Chinese versions and translations of Vinltadeva's commentary. 5. The Sanskrit for this verse has survived. It reads: yady aplndriyavijfiapteJ:! karaJ)am paramliI).avaJ:! atadabhataya nasya ak~avad vi~ayo 'IJavaJ:!. 6. Viikyapadfya 1:30-31:

na cagamad ne dharmas tarkeIJa vyavati~thate api yajjfianam tad apy agamapiirvakam

r~IIJam

dharmasya cavyavacchinnaJ:! panthano ye vyavasthitlil:J na tliIilllokaprasiddhatvat kaScit tarkeIJa badhate 7. Vakyapadfya 1:40:

idam pUIJyam idam papam ity etasmin padadvaye aCal)Qaiam manu~yliI).am aJpam S3straprayojanam 8. Vakyapadfya 1:36 and 1:38: pratyak~ anumanarh ca vyatikramya vyavasthitlil:J pitrrak~pisacanam karmaja eva siddhayal:t

atIndriyan asarilVedyam paSyanty ar~eIJa cak~u~ ye bhavan vacanam t~am nanumanena badhyate

Notes

218 9. Viikyapadfya 1:32-34:

avasthadesakalanam bhedad bhinnasu saktisu bhavanam anumanena prasiddhir atidurlabha

mm

arthakriyam prati nirjfiatasakter dravyasya t3rtl visistadravyasam bandhe sa saktil,! pratibadhyate yatnenanumito 'py arthal). kusalair anumatrbhil). abhiyuktatarair anyair anyathaivopapadyate 10. Patafijali 1880 ed., p. 1 11. The discussion occurs atPramiilJasamuccaya 5:17-18. See also 2:16. 12. The principal passage from which I am extracting this argument is PramiilJasamuccaya 5:18. As can be seen by looking at the translation of that passage, it is very compact and requires much expansion. There is, therefore, a need for caution in following my interpretation. 13. See PramiilJasamuccaya 2: 13. 14. This issue comes up for discussion at PramiilJasamuccaya 5:50. 15. See Cardona 1967:336-338 and 345-348. 16. See PramiilJasamuccaya 5:2ab: "na jatisabdo bhedanam anantyavyabhicaratal). vacaka/.l." In this statement "anantya" rules out a word's expressing the totality of particulars to which it applies, and "vyabhicara" rules out a word's picking out only part of its domain. 17. See PramiilJasamuccaya 5:2cd and Diimaga' s own commentary thereto. 18. A discussion of the arguments used by Dirmaga is to be found in Matilal1968. 19. See PramiilJasamuccaya 5:8cd and 9cd-lOa. 20. See PramiilJasamuccaya 5:9ab. 21. See PramiilJasamuccaya 5:4a and its commentary. 22. The particular problems of interpretation will be dealt with in my own commentary to the translation of PramiilJasamuccaya 5:4b-8b below in chapter 7.

23. PramiilJasamuccaya 5:4b. 24. My presentation here follows Iinendrabuddhi's analysis of PramiilJasamuccaya 5:7b.

Chapter Five 25. PramiifJ.asamuccaya 1:3d. See also the discllssion above, p. 134.

26. MatilaI1971:37.

219

TRANSLATIONS

Introduction to translation In the two chapters that follow I have presented English translations of passages selected from PramiilJasamuccaya chapters two and five. Chapter six of the present study contains the fIrst twenty-five verses (kiirikii) of the second chapter of the PramiilJasamuccaya, in which Diri.naga sets down his own views on the limits of the inferential process. Chapter seven contains four selections from PramiilJasamuccaya chapter fIve. The first selection comprises PramiilJasamuccaya 5:1-13, in which Diri.naga discusses the nature of the knowledge yielded by verbal symbols and the content of that knowledge. The second selection, consisting of PramiilJasamuccaya 5:14-16, deals with the question of how symbols that express preclusion.(apoha) combine to form longer units of expression, the relationships between longer expressions and their components, and the relations among the components within longer expressions. In the third selection, consisting of PramiilJasamuccaya 5: 17-20, Diri.naga deals with the relations among components in complex states of affairs consisting of individuals and their properties. He deals as well with the relations between symbol-complexes and property-complexes. In this context Diri.naga offers his main objections to the view that universals and qualities exist as realities that have a relation to individuals, which relation exists outside cognition as a fact to be discovered by awareness. The fourth selection contains PramiilJasamuccaya 5:25cd-3.6 (omitting 31d-32) and deals in general with the question of what it is exactly that a given term's application to a subject of discourse precludes. This in turn leads to a discussion of the metaphysical basis underlying the contrariety of terms and of propositions. To all of the passages translated I have added my own commentary, which not only contains information on how Diri.naga's arguments were interpreted by such philosophers as Uddyotakara, Jayamisra, Parthasarathimisra, and Jinendrabuddhi, but also ventures occasionally to express the line of arguments used by the classical Indians in a way intended to make them a little more accessible to modem readers. The principal text on which these translations are based is Kanakavarman's Tibetan translation of the PramiilJasamuccaya entitled Tshad ma kun las bdus pa. Only occasionally does Kanakavarman's translation seem less coherent than Vasudhararak~ita's, and on those

223

224

Introduction to translation

occasions I have followed the translation of Vasudhararaksita and have so indicated in notes. Since both of the two Tibetan transiations are very difficult to interpret, and since their interpretations of Diimaga are so frequently different from one another, I have relied heavily on the Tibetan translation of the subcommentary (([ka) by Iinendrabuddhi. More will be said below about the Tibetan translations of all these works, but first I should like to offer some account of how I have gone about doing this translation. First, on the question of sources, it cannot be emphasized too strongly how difficult of access Difmaga's thought is. His style of composition is generally procataleptic. That is, he presents his position and argues for it summarily, then strengthens his position by entertaining and ultimately rejecting any number of plausible counterpositions to this own. But he also refines his basic position by entertaining possible objections to it and modifying the first expressed position in light of those anticipated objections. It is in practice not always possible to determine, when confronted with a particular passage in the Pramar;asamuccaya, which sentences represent refinements that Dirinaga wishes to incorporate into his own theory and which sentences represent an.ticipated counterarguments to his position. Consequently, while it is most often clear what his overall position is on a matter, it is just as often unclear what his opinion is on many more subtle points within the confines of the overall opinion. Deciding these matters would most probably be a serious problem even if we had before us the original Sanskrit texts of Dmnaga's works. When what we in fact have are two Tibetan translations that disagree in their interpretations on numerous subtle points, it must be conceded that the obstacles standing in the way of our gaining a perfectly accurate picture of Dirinaga's thought are formidable. And therefore we have little choice but to tum to sources outside Dirinaga himself for clues to his views on many matters. The obvious question is, which outside sources are reliable guides to Dirinaga's thought? On first consideration it would seem reasonable to expect that DharmakIrti's work would be as reliable as any, for DharmakIrti is relatively near in time to Dirinaga. If Dirinaga's dates are ca. 480-540 C.E. and Dharmaldrti's ca. 600-660 C.E., then DharmakIrti is much nearer in time to Dirinaga than are any of the sources that I rely so heavily upon: Iinendrabuddhi (eighth century?), Pan:hasarathimisra (late eleventh century), Jayamisra (late seventh or early eighth century). Moreover, DharmakIrti is honoured by tradition as the expositor par excellence of Dirinaga's thought and is credited by legend to have understood Dirinaga far better than even Dirinaga's own direct disciple, Isvarasena. That we cannot place a great deal of confidence in Dharma-

Part Two

225

kIrti's interpretations of DiIinaga as guides to DiIinaga's own arguments is something that I have argued in Chapter one and will not repeat here. What does need to be justified here, however, is my reliance on authors who postdate DharmakIrti and who without any doubt were themselves influenced by DharmakIrti in their overall understanding of DiIinaga's thought. Especially in need of some clarification is how it is that I, knowing that Iinendrabuddhi believes that Dharmaklrti's interpretation of DiIinaga is essentially correct, trust Iinendrabuddhi any more than DhatmakIrti. One obvious feature of Iinendrabuddhi's commentary that makes it more useful as a guide to DiIinaga's thought than DharmakIrti's work is that Iinendrabuddhi offers an explanation of nearly every phrase in the Pramii1;lasamuccaya, for both the verses and the prose autocommentary. This accountability for every single passage of the work he is explaining compels him to formulate some hypothesis on what the overall strategy of a passage is and the tactics by which that strategy is carried out. That is, he must offer some account of what the passage aims to establish and how each sentence in the passage fits into the overall plan of the passage. Insofar as Iinendrabuddhi's commentary is concerned with this analysis of the strategy and tactics of the text commented upon, it is a most skillfully executed piece of work and generally inspires full confidence in its faithfulness to the theories of the text it is trying, sometimes with almost as much difficulty as a modern scholar feels, to explain. DharmakIrti, on the other hand, is less interested in offering a detailed commentary on the passages of DiIinaga's work than in building his own structure and drawing upon the authority of quotations from DiIinaga almost as "proof texts. " Even when Iinendrabuddhi leaves off the task of phrase-by-phrase commentary and introduces further arguments of his own in support of DiIinaga's conclusions, the deviations from DiIinaga's conceptual framework are easy to spot and cause little confusion. 1 Thus, for example, when Iinendrabuddhi brings the concept of causal efficiency (arthakriyii) into the discussion of what differentiates real entities from conceptual structuring, it is easy to see that the explanation is gratuitous in that it goes entirely beyond any explanation that DiIinaga himself explicitly states or even suggests. Similarly, when Iinendrabuddhi accounts for certain speech habits and mental attitudes as speaker's predispositions inherited from the karmic influences of previous lives, it is not difficult to see that ideas very different from any that DiIinaga himself appeals to are being called into service. 2 But these deviations from DiIinaga's own argumentation do not flaw Iinendrabuddhi's commentary, for they have no serious effect on the accuracy of the phrase-by-phrase account of DiIinaga's writing. Rather,

226

Introduction to translation

these supplementary argument~ and explanations that Jinendrabuddhi provides may be seen as attempts to defend Difmaga from his critics by pulling him into line with what had come to be the orthodox interpretation of the system built upon the foundation of his thought. For this reason I have all but ignored Dharmakrrti's views on Diimaga, feeling that Dharmaldrti's work is a study unto itself that is best kept separate for the time being from any study of Dinnaga, and have felt confident that in following Jinendrabuddhi I would not be led too far into the wilderness. Negotiating the text by my own wits alone without following Jinendrabuddhi's guidance, I would surely have made countless outrageous blunders of interpretation; as the translation now stands, I hope that the many imperfections in it are not outrageous but merely a little comical. The Pramii1).asamuccaya is composed in mixed verse and prose in a style in which verse passages are so interwoven into the prose passages that neither the verse text nor the prose text stands alone as an intelligible text. In my translation I have separated the verses from the prose commentary, setting them off in large letters. The verses are numbered KO.O, where "K" stands for the number of the kiirikii as given in the critical edition by Hattori Masaaki (Dinnaga 1982 ed.). Following each verse is the prose commentary, in which the parts of the verse being explained are embedded; these embedded parts of verses are indicated by italics, and the prose passages are numbered KY.O, where "V" stands for the number that I have assigned to a passage of prose commentary under a given verse. My own commentary is keyed to the translations of Dinnaga' s commentary in passages numbered KV.T, where "T" is a number I have assigned to passages to my commentary. Notes that contain information about key Sanskrit or Tibetan words are placed at the end of each chapter.

The history of the Pramii1J.asamuccaya in Tibet Dinnaga's Pramii1).asamuccaya is known to have been translated into Tibetan three times. Bu ston reports that a pundit by the name of Candrarahula was invited to Tibet and that he and the Tibetan translator (10 tsii ba) Ting nge 'dzin bzang po translated the Pramii1).asamuccaya and other works. 3 Since Candrarahula is named, along with Atlsa (982-1054), as one of the teachers of 'Gos khug pa, a disciple of the great scholar 'Brog mi (992-1074), it is likely that this early translation of the Pramii1).asamuccaya was executed in the middle part of the eleventh century.4 The

Part Two

227

index to the Peking edition of the Tibetan Tripitaka lists eight translations by Ting nge 'dzin bzang po, but his translation of the PramalJasamuccaya does not survive in the Peking or in any of the other redactions of the . Bstan 'gyur. One of the two Tibetan translations of the PramalJasamuccaya that still survive was done by a layman (dge bsnyen; upasaka) by the name of Zha rna Seng ge, who is referred to by the names of Seng ge rgyal po, Seng ge rgyal mtshan or simply Seng rgyal. According to Bu ston, Zha rna Seng ge executed his translation in collaboration with several pundits, but the only one named by Bu ston is one whose name in Tibetan translation is Nor bzang srung ba, which probably represents the Sanskrit name MaJ)ibhadrarak1;lita. 5 Zha rna Seng ge's translation still survives in the Sde dge, Co ne, Peking and Snar thang editions of the Bstan 'gyur. The collophon to this translation gives both the Tibetan form, Nor 'dzin srung ba, and the Sanskrit form, Vasudhararak1;lita, of the Indian pundit with whom Zha rna Seng ge worked. It is recorded that Zha rna Seng ge was the youngest of six children of Zha ma rdo Ije rgyal mtshan, three of whose offspring became prominent scholars. Since Zha rna Seng ge studied the art of translation under Rma 10 tsa ba (b. 1044) and Rngog 10 tsa ba (1059-1109), it seems most likely that he translated the PramalJasamuccaya sometime in the late eleventh or early twelfth centuries, a time of great moral reform and renewed scholastic zeal among the Buddhists of Tibet. Zha rna Seng ge is reported to have been a master of several fields of Buddhist literature, including exegesis, Madhyamaka philosophy, abhidharma and logic, and is credited with the translation of Maitreyanatha's Dharmadharmatavibhangaktirikti and several Madhyamaka works and commentaries to the Prajnaparamita and tantric writings. Despite Zha rna Seng ge's education, his reported intellectual prowess and his general skill as a translator, his rendering of Dinnaga's work is far from perfect. It tends to be unnecessarily literal and mechanical and often shows poor judgement in rendering technical terms into Tibetan. Also since Tibetan is much less tolerant of compounding than Sanskrit and the Indo-European languages in general, a Tibetan translator is often compelled to make explicit in his translation the syntactic relations of words that appear in long compounds in the original Sanskrit. In the analysis of compounds Zha rna Seng ge's translation is particularly unreliable, often giving them a misleading interpretation. I suspect that these imperfections in Zha rna Seng ge's translation may be due largely to his pundits' imperfect understanding of Dinnaga's writing. Neither Vasudhararak1;lita nor MaI)ibhadrarak~ita is mentioned in 'Gos 10 tsa ba Gzhon nu dpal's Blue Annals, nor does either name appear in either its Sanskrit or

228

Introduction to translation

its Tibetan form on any text in the Tibetan Tripiiaka other' than the PramiilJasamuccaya, so it is possible that Zha rna Seng ge had to make do

with less than distinguished pundits on this project. And it is also possible that neither he nor his pundits had access to Iinendrabuddhi's subcommentary, without which much of Dinnaga's work is almost hopelessly difficult to decipher. Indeed, many of the errors in translation made by Zha rna Seng ge are of the sort that would probably have been avoided had he been thoroughly familiar with Iinendrabuddhi's commentary. The second of the two SUrvIVIng translations of the Pramar:wsamuccaya was done by the Tibetan Dad pa'i shes rab in collaboration with an Indian pundit named Kanakavarman (Oser kyi go cha). Dad pa'i shes rab's name does not appear on any other text in the Tibetan Tripitaka, and neither he nor Kanakavarman is mentioned in 'Oos 10 tsa ba's Blue Annals. In the Peking redaction of the Bstan 'gyur Kanakavarman's name appears altogether on twelve texts--on six as a principal translator and on another six as a reviser of earlier translations done by others. As a reviser he worked most often with Pa tshab 10 tsa ba Nyi rna grags pa, another very active translator and reviser who is mentioned frequently in the Blue Annals as a great scholar who trained numerous other translators. His period of activity was the early twelfth century. Therefore, Kanakavarman's translation of the Pramar;asamuccaya was probably done at approximately the same time as Vasudhararak!iita's.6 Kanakavarman's translation of Dinnaga's work is generally speaking far more clear than the translation by Vasudhararak~ita and Zha rna Seng ge, and it is also more in line with the grammatical analyses arid philosophical expositions given by Iinendrabuddhi. In cases where a fragment of the original Sanskrit passage has survived, a comparison of this fragment with the two Tibetan translations most often shows that Kanakavarman's shows more finesse and accuracy than the other translation. It is surprising, given the general superiority of Kanakavarman's translation, that it does not appear in the Sde dge redaction of the Bstan 'gyur, especially since all of Kanakavarman's other translations do appear there. Iinendrabuddhi's Pramar;asamuccayatfka entitled Visiilamalavatf was translated into Tibetan by a monk (dge slong) named Dpalldan Blo gros brtan pa (1276-l342), a very productive scholar whose name appears as translator of twenty-six treatises in the Tibetan Bstan 'gyur. Twelve of his translations were of tantric texts, and fourteen were of scholastic treatises. Aside from Iinendrabuddhi's commentary to Dmnaga's

Part Two

229

the only other text dealing with logic and epistemology that Blo gros brtan pa translated was the Tarkabhiisa of Moksakaragupta, a 'compendious text that summarizes the conclusions reached at different stages of the six centuries of evolution of logical and epistemological theory by Indian philosophers. But the great majority of BIo gros brtan pa's translations were of works on Sanskrit grammar, poetics and lexicography. Of the twenty-eight texts dealing with linguistic science (sgra rig pa; *sabdavidya) that are preserved in the Bstan 'gyur, Blo gros brtan pa translated eight. Indeed his contributions to the progress of the study of technical aspects of Sanskrit grammar are rivaled only by those of his contemporary Nyi rna rgyal mtshan dpal bzang po, the Thar pa 10 tsa ba who translated several works of the Buddhist grammarian Candragomin and who taught Sanskrit to Bu ston rin po che, the great polymath and redactor of the Tibetan Tripitaka. Given this great interest in Sanskrit grammar in the early fourteenth century Tibet, it seems likely that Jinendrabuddhi's commentary to Diimaga was translated at this time in the belief that Diimaga's commentator was the grammarian Iinendrabuddhi who wrote a commentary on the Kasikavrtti. Blo gros brtan pa's knowledge of Sanskrit works on grammar (vyakaraf}.a) made him particularly well qualified to translate the Visi'ilamalavatf of Jinendrabuddhi, which is rich in technical discussions of grammar. Especially for the fifth chapter of the Pramaf}.asamuccaya in which the main topics are linguistic, Iinendrabuddhi's expertise in grammar sheds much light on DiIinaga's presentation, and Blo gros brtan pa's solid grounding in Sanskrit helped him in turn to produce a clear and accurate translation of Iinendrabuddhi. This translation of Jinendrabuddhi' s Visalamalavatz is probably the most important study in the Tibetan language on the philosophy of DiIinaga. But Iinendrabuddhi's expertise in grammar also makes his commentary a rather difficult text to read in Tibetan, a language in which the technical terms of Sanskrit grammar become distorted almost beyond recognition. Indeed, anyone who might try to read DiIinaga or Iinendrabuddhi in Tibetan without first gaining a good grounding in Sanskrit linguistic vocabulary would likely come away with the feeling that many of the passages of these two texts contained little more than a random juxtaposition of words, like a dhiiraf}.z. PramafJ:asamuccaya,

Some pas ages of my own English translation of DiIinaga may seem hardly much easier to read than the original Sanskrit fragments or the Tibetan translations. No one is more painfully aware of this fact than 1. It represents the struggle of many years of work and a considerable amount of reflection. It has been reworked many times over in an attempt to make it more clear and more consistent, but it is still quite an ordeal to read through it with any kind of understanding. It is for this reason that I have

230

Introduction to translation

seen fit to provide a running commentary in which I have followed the time-honoured commentarial practice of explaining the obvious points in great detail and passing over difficult points in silence. We must, I believe, face the simple truth that a great deal of Diimaga's work has been irretrievably lost, except in the unlikely event that someone recover the original Sanskrit versions of his key works. Trying to piece together his thought on the basis of the Tibetan translations is like looking at a human skull and trying to imagine what the person's face looked like when alive. We shall probably never know for certain what many of the details of Dinnaga's philosophy were all about, but we are in a good position to hazard a few conjectures. It is my hope that other scholars will look at this translation from time to time and bring their own expertise in closely related areas to bear and that eventually, by a great cooperative effort, the scholarly community will someday possess a much better picture of Dinnaga's system of philosophy than is found in this meagre sketch .

•- Notes --

l. See, for example, JP285a-288b. Stcherbatsky (1932:461-470) translated this passage into English.

2. See, for example, the passage beginning at JP352b. 3. Obermiller 1932:215. 4. Roerich 1949:360. 5. Obermiller 1932:221. 6. I am grateful to D. Seyfort Ruegg for pointing out to me that Nyi rna grags pa is listed in the index to the Blue Annals under the name Pa tshab 10 tsa ba. In my Ph.D. dissertation I had incorrectly stated that Nyi rna grags pa was unknown to the author of the Blue Annals and must therefore have flourished after 1478, the date of the composition of the Annals. Hattori Masaaki (1968:13) also suggested, following the same reasoning, that Kanakavarman worked in the late fIfteenth century.

Chapter 6 PramiifJasamuccaya II: "On reasoning" 6.1 Inference differentiated from sensation

1.0.0 The inferential process is of two kinds: that which is for one's own sake, and that which is for the sake of other people. Inference for oneself consists in discerning an object through a sign that has three characteristics. As was the case above, this too refers to the resulting cognition. Their fields of operation and essential natures are dissimilar. 1.1.0 "The inferential process is of two kinds: that which is for one's own sake, and that which is for the sake of other people." Of those two "inference for oneself consists in discerning an object through a sign that has three characteristics." Inference for oneself is discerning an inferable object through a sign that has the three characteristics explained below. 1.1.1 The term "sign (Ungar refers to a property that serves as evidence for another property that shares the sign's locus. Thus the term "sign" is virtually interchangeable with the term "evidence (hetu)."

i.2.0 "As was the case above, this too refers [not only to the cognitive process but also) to the resulting cognition." The resulting cognition is explained in this case in the same way as it was explained in the case of sensation, that is, with reference to a cognition's having two aspects. 1.2.1 The reference in this passage is to Pramiir,zasamuccaya 1:8-12. Like the English word "inference" the Sanskrit word "anumiina" has two distinct meanings: a) the process of inferring, and b) the cognition that results from that process. Similarly

231

232

PramafJasamuccaya 2: 1

On reasoning

the Sanskrit tenn "pramal}a" is taken to refer both to the process of cognizing and to the resulting cognition.

1.3.0 Now if both [sensation and inference] are characterized as cognitions, what is the difference between them? "Their fields of operation and essential natures are dissimilar." Sensation and inference have distinct fields of operation, and their essential natures are also distinct in accordance with their having different cognitive images. 1.3.1 The field of operation (vi~aya; yui) of a given cognitive process is that set of objects that are knowable through that process. Iinendrabuddhi (P94~) explains that sensation has external objects making up its field of operation, as consequence of which its cognitive images are vivid (snang ba gsai ba; *pratibhasa). But the objects of inference are not external objects but concepts, so inference's cognitive images are vague.

a

2.0.0 Because the particular is inexpressible. [Inference], since the object grasped by it differs, is otherwise. It is described through the fact of its being seen, but it is not named through its essential property. 2.1.0 Now why is it that only inference is subdivided into two parts? "Because the particular is inexpressible. [Inference}, since the object grasped by it differs, is otherwise." Sensation and inference have different fields of operation. If the object of sensation were expressible, one could infer it just through speech. 2.1.1 In Diimaga's system "inference" refers to a) the acquisition of new knowledge through reasoning, and b) the communication of what one knows through argument or discourse. The knowledge of a particular, however, which is vivid and exact, cannot be transmitted verbally, since verbal communication is necessarily vague and inexact. Therefore, "sensation" refers only to the acquisition of new knowledge, never to the communication of what one has sensed. Consequently it cannot be divided, as inference is, into sensation for oneself and sensation for others.

2.2.0 Now one might argue that sometimes we observe cases of inferring sensible objects, as for example (when we infer) a tangible property through a visible one. It is true that there are such experiences, but it is not [really a case of inference and sensation having the same field of operation]. The inference of that [tangible property] is not the same as the sensation of it, but rather it is otherwise; it is otherwise in that we infer a universal of the tangible property after recalling a former experience. Thus we infer the tangible property through the universal of the visible property rather than through the form of the sensation of the visible property itself. Since the particular tangible property that was previously sensed cannot be

Chapter Six

Prami'iT)asamuccaya 2:2

233

designated by name, there is no confusion of the fields of operation of the two means of acquiring knowledge. 2.3.0 Now if sensible things are inexpressible, why are expressions such as "seen" and so on used with reference to things that are seen and so forth? There is no inconsistency here, for in that case "it is described through the fact of its being seen, but it is not named through its essential property." They are referred to by some token such as "is seen" "is heard" "is desired" "is known" and so forth but not through their essential prope:rties.

3.0.0 Since it is cognized through a name it is the cognitive image which is different that is expressible. The mind has two cognitive images. Is [inference] observed even when there is no universal? No, for it is a universal that is indicated. 3.1.0 But is it not the case that after we apprehend a blue colour through apperceptive awareness, the very object that we experienced through visual awareness is then expressed thus: "It is blue"? In this case too, "since it is cognized through a name it is the [universal] cognitive image which is different [from the cognitive image of the particular] that is expressible. The mind has [the capacity of grasping] two cognitive images." What this says is that since apperceptive awareness, by rejecting what is not blue, is able to receive the object experienced by visual awareness, the mind has two cognitive images [that is, the image of the universal and that of the particular]. 3.1.1 The question that is brought up at the beginning of this passage arises from the perspective of Buddhist abhidharma. According to the analysis of experience offered there, human beings have six sense faculties, namely, the five external faculties of seeing, hearing, smelling, tasting and feeling, plus the internal faculty of thinking. Each of these faculties has a discrete field of operation, which comprises all the sense data that can be apprehended by the sense faculty in question. The five external fields of operation are those of sights, sounds, odours, tastes, and touch, and the internal field is that of thoughts. When a sense datum stimulates a functioning sense faculty of the type that can sense the type of datum in question, the result is awareness. Given that there are six distinct types of sense faculty, there are also six distinct types of awareness. The five sorts of external awareness are visual awareness, aural awareness, olfactory awareness, palatal awareness and b04ily awareness, and the internal type of awareness is mental awareness, which may also be called apperceptive awareness, because according to the abhidharma system the faculty of thinking can take as its data thoughts about the data that the other sense faculties bring to our awareness. It is, therefore, a second-order awareness, an awareness of awareness. Hence after visual awareness, which consists in the experience of some sight such as a patch of blue colour, there arises apperceptive awareness wherein the faculty of thinking applies a concept or name to

234

Pramil/Jasamuccaya 2:3

On reasoning

that blue object. What Diimaga calls sensation (pratyalqa) seems to comprise exactly the five types of extelnal awareness discussed by the followers of abhidharma, while what he calls structuring (kaipana) appears to correspond fairly closely to the abhidharma category of apperceptive awareness. So, given this analysis from the perspective of abhidharma, it would appear to be the case that a particular sense datum such as a patch of blue can be the object both of sensation and of thinking; if so, it would appear that Diimaga is incorrect in saying that the fields of operation of the external senses are quite distinct from the field of operation of thinking or the formation of concepts. This same question from the a~hidharma perspective comes up also in Diimaga's own commentary to PramalJasamuccaya 1:4. 3.1.2 In the above passage Diimaga says "since apperceptive awareness, by rejecting what is not blue, is able to receive the object experienced by visual awareness, the mind has two cognitive images." In his commentary on this passage Jinendrabuddhi (P96a!) says: "There is a cognitive image of the universal that is different from the cognitive image of the particular. It is by means of that universal cognitive image that a thing is expressible, not by means of the particular. Its name is the term 'blue,' but the object cognized [through that word] is definitely not the particular. Therefore this verse asserts 'the mind has two cognitive images.' The phrase 'by rejecting what is not blue' indicates the other cognitive image. What this means is 'by excluding non-blue from the object that those who are expressing it in language are thinking about.'" Before each person speaks, her potential audience can imagine anything whatsoever as that about which the speaker is thinking. With each word that the speaker utters to express her thoughts, however, the audience is obliged to eliminate certain things from the universe of discourse, namely all those things that are logically incompatible with the "meaning" of the words spoken. After the speech is over, the audience still does not know precisely what the speaker was thinking, but the audience does know at least some of what the speaker was not thinking. Apperceptive awareness, according to DiI'maga, classifies its data in essentially the same way as an audience assimilates what a speaker has said--by grouping together under a rough class-heading those data of experience that are not mutually incompatible. This mental activity, unlike sensation which involves the relatively passive reception of data, is one that involves the relatively active manipulation of received data, or rather of concepts formed on the basis of received taw data.

3.2.0 Given that the particular is inexpressible, since it has the essential nature of a knowable object in the field of operation of sensation, whereas inference has universals as its field and is expressible through verbal expressions, inference alone is subdivided into two kinds. But, it may be objected, we should not say that all inference has universals as its field of operation, because "it is observed even when there is no universal." Although [the element] wind does not have the character of a universal, it is nevertheless observed to be inferred by means of its touch, as it is said "Although it is sensible through touch, its touch is not that of things that are seen" and so forth.! "No, that is not the case,for it is a universal that is indicated." It is not a case of inferring [a particular sUDstance, namely the element] wind, because since touch and so forth are qualities, the

Chapter Six

PramalJasamuccaya 2:3

235

general property of having a substratum is indicated [indirectly through the quality-universal). Or, to explain it another way, it is not the specific nature of wind and so on that is inferred, but it is just the fact of being supported by some substance, which fact is common to touch [and the other qualities], that is indicated. 3.2.1 According to the system of categories taught in the Vaise~ika school, there are nine basic Substances (dravya): earth (prthivl) , water (apas), fire (tejas) , wind (vayu) , ether (akiisa), time (kiila), space (dis), self (atman) and mind (manas). Inhering in those substances are various qualities (gUlJa). The system recognizes a total of seventeen qualities, and it is said that each substance can be known by the qualities that inhere in it. Now according to this system, the quality of touch is the only sensible quality that inheres in the substance wind. Whereas earth has the sensible qualities of colour (rupa) , taste (rasa), smell (gandha) and touch (sparsa), and water has colour, taste and touch, and fIre has colour and touch, wind has only touch. But its touch is of a special kind that differs from that of the other substances. Accordingly, when one has a particular sensation of touch, one can infer the presence of the particular substance wind. In this passage and the passages that follow Diimaga argues against various aspects of this Vaise~ika doctrine. In this passage he argues that by perceiving the quality touch one can infer only that there is some substance in which it inheres, since a quality must inhere in a substance; but beyond this general fact, says Diimaga, one can conclude nothing. He evidently rejects the doctrine that wind has a particular kind of touch that is unique to it.

3.3.0 Suppose it is argued that it [scilicet, the particular kind of touch that is putatively unique to the substance wind] is confirmed by a process of elimination. 2 That is, suppose it is established that this inference regarding the nature of things such as wind is through a process of elimination, as follows: "The touch is absent in visible things, but not in invisible things." That will not do, because it has not been confirmed as existing, and because the touch could be denied in the same way. While it is not possible to disconfirm its presence, you are confmning wind and so on by inferring a specific substance through a specific quality by excluding a general quality from a relation with the specific quality. Its presence, however, is not confirmed and its rejection is the same as with earth and so forth, because there is no special quality.3 3.3.1 As we have seen above (3.2.1), it was the contention of the Vaise~ika scholastics that the species of touch that occurs in the substance wind is different from that touch which occurs in the visible substances. Ditmaga challenges this contention directly by saying that it has not been established that the touch that occurs in wind is indeed discernible from the touch that occurs in the visible substances. If it is possible to deny that this putatively special kind of touch occurs in earth, water and fire, then there is no apparent reason why we should not also deny that it occurs in wind. Diimaga concedes that it is not possible to disprove the Vaisesika claim, but at the same time he contends that the burden of proof is on them to show why the tactile properties of wind should be regarded as distinct from the tactile properties that occur generally in the other three elements.

236

PramalJasamuccaya 2:3

On reasoning

3.4.0 Suppose one argues that touch is denied in visible substances just on the ground that it is invisible itself. That also is incorrect. Touch cannot be denied as a quality of things that are visible, corporeal and resistant. In fact, the mind infers touch as a quality of those things because touch is observed when [those] other [properties] are observed, and it is not observed when they are not observed. Therefore touch does belong to visible things; Thus, since there are more possibilities than one, confusion arises as to what touch does belong to, so one cannot infer wind by denying all other possibilities. 3.4.1 Some substances, such as ether and self, are regarded to be ubiquitous, so obviously several ubiquitous substances can occupy the same space. In contrast to these ubiquitous substances, some substances have the property of resistance (pratigha; thogs pa) whereby they exclude other substances that also have the property of resistance from simultaneously occupying their space. Now DiI'maga argues here that objects have this property of resistance if and only if they are both tangible and visible; therefore, the property touch belongs only to substances that have the property of being visible, so it cannot belong to invisible wind as the Vaise~ikas suggest.

4.0.0 Suppose it is confirmed because there is no difference, because it is by denying what is other than it on the authority of the statements of credible persons. That is not the case, because in fact they are different. 4.1.0 Moreover, [the particular kind of touch that is putatively unique to wind is not learned through inference] "because it is [confirmed} by denying [substances} other than it on the authority of the statements of credible persons." In this case, Wind is established after one first infers substance in general and then eliminates the other substances on the authority of the statements of credible persons. The same applies to other cases as well. Therefore, it is on the basis of scripture (agama) that wind is confirmed [as the locus of a special kind of touch]. "Suppose it is confirmed because there is no difference." Since there is no difference between scripture and inference, it is established here that inference has a particular as its field of operation. 4.1.1 The credible persons (apta; yid ches) referred to in this passage are the people whose statements the followers of the Vaise~ika school believe. It is the statements of these persons, who are deemed credible because they are known to have both competence and integrity, that make up the scriptures (agama) that the orthodox Brahmanical philosophers follow. Now Ditmaga is committed to the view that in the final analysis scripture is no different in principle from inference. Therefore, anything that is established on the basis of scriptural authority is, in Dlfmaga's view, in the final analysis actually based on inference. Suppose it has to be conceded to the Buddhists that it is because of the statements of these authorities that the

Chapter Six

Pramiil).asamuccaya 2:4

237

Vaise~ikas are in a position to know the features peculiar to the various substances. But if this is the case, then when scripture deals with what is particular and unique, it ought to be acceptable to Diimaga as an instance of inference dealing with what is particular. Therefore, Diimaga's original contention that inference and sensation deal with quite distinct fields of operation is apparently falsified by this counterexample. Diimaga entertains this anticipated argument against his position and responds.

4.2.0 "That is not the case, because in/act they are different." That verbal testimony (siibda; sgra las byung baY is different from inference is confirmed on the authority of common usage (riiq.hi, grags pay. Some say the distinction lies in the fact that [in verbal testimony] no observed precedent (drS{iinta) is stated. If this were so, it would follow that whenever one or both observed precedents were not explicitly stated [in a formal debate] because they are already well known, that would be [a case of] verbal testimony [rather than inference]. But others state the difference as follows: in the case of verbal testimony, the word indicates [its object] owing to an identity between the word and the object, so in fact it does not reveal the object through a process of inference. But if this is the case, that a word signifies through identity with the word, some account must be given of how we arrive at a conclusion of the object. The object named by the word "tree" is nothing other [than the tree}. Although the word "substance" makes the same object known, it really does so in another way, by distinguishing it from what has no substance. Now one might argue that the word "tree" also makes the distinction from non-substance known. True, it does make that known, but it does so by implication, not explicitly, so that objection is invalid. Now there may be nothing wrong in saying that verbal testimony is a means of acquiring knowledge in the case of such words as "tree," but words such as those that name actions also make us know something, but not through some characteristic. Some assert that the only speech is the [entire] sentence (viikya, ngag) and the individual word (pada; tshig) is a means of understanding (buddhyupiiya; rtogs pa'i thabs) that sentence. They do not recognize an object conforming to speech. Thus, admitting a slight difference between inference and verbal testimony, they say that the two are different. In the above explanations of inference, attention has been focused on signs not connected with speech. But verbal communication also does not apply to a unique thing. Therefore, one should regard inference as being of two types according as [its object is] visible or invisible. In the case of a visible object, we may teach its name. With reference to an

238

Pramar;asamuccaya 2:4

On reasoning

invisible object there is only a concept (vikalpa; rnam par rtog pay but there is no awareness of a particular object. 4.2.1 While Dirinaga maintains that in the final analysis there is no distinction between the process of acquiring information through reasoning and the process of obtaining information through scripture, he also holds that there is in ordinary language a distinction drawn between the two processes. There are a" number of different views on how exactly an inference differs from an appeal to verbal testimony or the authority of scriptures, but what they all have in common is that a distinction is to be made between the interpretation of inferential signs and the interpretation of linguistic signs. 4.2.2 "Some assert that the only speech is the [entire] sentence and the individual word is a means of understanding that sentence." This is the view that Dirinaga himself accepts. He defends this view at Pramiil'Jasamuccaya 5:46-49, acknowledginf, it to be the view of Bhartrhari. The issue here is not, as Vasudhararak5ita (29b ) and Kitagawa (1965:90) have it, that the individual word is incapable of making its object known, but rather as Kanakavarman and Iinendrabuddhi (p102a2) have it, that words make the sentence known, which in turn conveys its meaning. In this view, individual words are useful fictions, conceptual entities arrived at through the process of abstraction, that can help us understand the meaning of a sentence when, for example, our command over a language is insufficient to enable us to grasp the meaning of a sentence immediately on hearing it. 4 4.2.3

"In the case of a visible object, we may teach its name. With reference to an invisible object there is only a concept but there is no awareness of a particular object." Iinendrabuddhi (p105a8) explains that in the process of learning an object's name by having the object pointed out while its name is spoken, we simultaneously grasp the object's peculiar attributes (svalakyal'Ja) and its general attributes (siimiinyalak~al'JG) and when that name is used later, then only the general attributes are communicated.

5.0.0 The statements of credible persons are inference insofar as they have the common character of not being false. Present in the object of inference and in what is similar to it, and absent in their absence. 5 5.1.0 How can verbal testimony be classed as inference? Words such as "heaven" do not express any object at all. "The statements of credible persons are inference insofar as they have the common character of not being false." Because when one hears the statements of credible people, the [resultant] cognition is not false, and because this makes them similar [to inference] we say they are inference. Furthermore, it is claimed that the name-giving was previously seen first hand. This view denies inference

Chapter Six

Pramiif}.asamucccrya 2:5

239

with respect to such things as the hypothesis of primordial substance (pradhiina; gtso bo).6 5.1.1 According to Ditmaga, inference is one of the two means of acquiring knowledge, and scriptural authority is subsumed under inference. This would seem to imply that scripture is therefore a means of acquiring knowledge. But the scriptures discuss a number of things that do not really exist, such as heaven and hell, so how can scriptures be credited with providing us with knowledge? This question could easily come from the perspective of a follower of the train of thought presented in Vasubandhu's Vijiiaptimiitrattisiddhi, in which it is argued that the hells cannot really exist as any kind of objective reality and must be understood simply as subjective experiences. The reply that Ditmaga gives is that not all scripture is to be regarded as a source of knowledge simply in virtue of its being scripture. Rather, a passage of scripture can be regarded as authoritative only insofar as what it says is true. So, rather than saying that something is true on the authority of scripture, we can only say that something in scripture has authority only if it is already known to be true. What this says in effect is that the truth of a passage of scripture must be confIrmed by some canon independent of scripture itself, and so scripture itself is not really authoritative at all. The question that this raises is: what standards are we to use in determining the truth of statements in scripture? The answer that Ditmaga gives is that we can use only the standards of experience and reason. Of these two standards, experience is the more powerful in that it alone can provide us with entirely new knowledge, whereas reason is limited to matters that have already been learned through experience. Said in another way, experience can positively increase our knowledge, but reason can only help eliminate some of our false beliefs. If we apply this observation to some claims that are made in scripture, we fInd that scripture often discusses things that may lie outside the realm of our own experience but could nevertheless be part of someone else's experience. But other claims are made that could not possibly have ever been part of anyone's experience, for they are utterly inexperienceable. One example of such a claim is the fundamental metaphysical doctrine of the Samkhya school, namely, that all physical as well as sentient objects in the universe are nothing but modifications in form of a primordial substance called "pradhana" or "prakrti". This primordial substance is in principle impossible to experience frrst hand, for it is in its primordial form only insofar as it has not taken on either one of the specifrc evolutionary forms that is capable of having a subjective experience of anything or one of the sensible objects that is capable of being experienced. But if it can never have been experienced, it can also never be the subject of a legitimate inference, for reasons that will become more clear as we look at the nature of inferential knowledge that is acquired through the interpretation of a proper inferential sign.

6.2 The three characteristics of legitimate evidence 5.2.0 The phrase "through a sign that has three characteristics" must be explained. [The sign is] "present in the inferable object and what is similar to it and absent in their absence." The inferable object is a property-bearer

240

Pram[l1J,asamuccaya 2:5

On reasoning

qualified by a property. After observing [the sign] there, either through sensation or through inference, one confirms that it is also present in a general way, either wholly or partially, in what is of the same class.7 Why is that? Because the restriction is such that [the sign] occur in only what is similar, there is no restriction that it only occur. But in that case it could be argued that nothing is accomplished by saying "it is absent in their absence." This statement is made in order to emphasize that [the sign], being absent in the absence [of subjectlike objects] is not present in what is other than or incompatible with the inferable object. Here then is the sign (liizga; rtags) with three characteristics from which we discern the signbearer (lingin; rtags can). 5.2.1 This passage is discussed in some detail above in chapter four, paragraph 4.2.2.2, so there is no need to say anything further here.

6.0.0 It is taken for granted here that there is also knowledge [of the sign],8 Because it is the principal one of the factors that produced knowledge [of itself]. It is of course the case that a sign having only one or two does not serve the purpose. 6.1.0 Should one not also mention in that case the knowledge [of the sign]? That is not necessary to mention, because "it is taken for granted here that there is also knowledge [of the sign]." How can that which is not explicitly mentioned be taken for granted? "Because it [scil., the sign] is the principal one of the factors that produced knowledge [of the si#n]." The sign is the foremost of the factors that make the sign known, and although awareness of the sign is itself dependent on one who is awate, it is nevertheless not dependent on many things such as an instrumental cause and so forth, so it is confirmed [automatically]. 6.1.1 The idea here is that the sign and the person who is aware of it together constitute the sufficient condition for awareness of the sign. If this were not so, that is, if a variety of other conditions were also necessa,ry to produce awareness of a sign, then the absence of anyone of these conditions would mean the absence of awareness of the sign. In such circumstances one could not take the awareness of the sign for granted just on the grounds of the sign's presence. But in fact, argues Diimaga, no other such conditions are necessary. Following a general principle whereby stating the sufficient causes of a thing is as good as mentioning the thing itself, Diimaga concludes that the presence of awareness of a sign, which awareness is a key element in inference, goes without saying once one has mentioned the sign itself. It is noteworthy that in most discussions of matters of logic and epistemology in classical Indian philosophy, psychological issues are never far in the background and are often brought into the foreground for special attention. In contrast to some trends in modem Western thought, where there has been a concentrated effort on the part of some to avoid psychologism,1O the classical Indians were relatively

Chapter Six

PramtilJasamuccaya 2:6

241

unconcerned with drawing careful boundaries between purely logical and purely psychological questions.

6.2.0 Since we have said that a sign has three characteristics, "it is of course the case that a sign having only one or two [of those characteristics} does not serve the purpose." Of these, signs having only one of the characteristics are as follows:

1. those which are present only in the subject of inference but are absent in what is similar and not absent in what is not similar, 2. those which are present in what is similar to the subject of inference but absent in the subject of inference itself and also not absent in what is not similar to it, and 3. those which are absent in what is dissimilar to the subject of inference but absent in the subject of inference and also absent in what is similar to it. Signs having only two of the characteristics are: 4. those that are present in the subject of inference and present in what is similar to it but not absent in what is dissimilar, 5. those which are present in the subject of inference and absent in what is dissimilar from it but are absent in what is similar, and 6. those which are present in what is similar to the subject of inference and absent in what is dissimilar but are absent in the subject of inference itself. The above six types of quasi-evidence (hetvabhasa; gtan tshigs ltar snang ba) can be understood by implication to be ruled out. Examples [of each of the six types of quasi-evidence are]:

7.0.0 Sound is eternal 1) because it is created, 2) because it is corporeal, 3) because it is unknowable. Sound is transitory 4) because it is incorporeal, 5) because it is audible, 6) because it is visible. 11 7.0.1

The six examples given in this verse represent every possible form of quasievidence discussed in the H etucakraniTlJaya, as well as three types of quasievidence not covered there. The first example, "Sound is eternal because it is created," represents situations 4 and 6 on the cycle of reasons, in which neither the association relation nor the dissociation relation is in place and so the sign indicates tlte contradictory of the property to be confirmed. The fourth example, "Sound is transitory, because it is incorporeal," represents situations 1,3,7 and 9 of the cycle of reasons, in which the evidence is associated both with the property to be confirmed and its contradictory and so is'inconclusive. The fifth example, "Sound is

242

Prama!)asamuccaya 2:7

On reasoning

transitory, because it is audible," represents the fifth situation in the cycle of reasons, in which the sign does not occur at all in the induction domain and so is inconclusive. The remaining three examples all have in common that they pertain to properties that do not have the first characteristic of proper evidence, namely, the characteristic of being a property of the subject of the inference, for sound is neither corporeal, unknowable nor visible. Quasi-evidence of this latter sort is not discussed in the HetucakraniTTJaya, which deals only with signs in which the first characteristic is in place.

6.3 Property·bearer as the subject of inference

8.0.0 Some people claim that it is another property that is cognized through the sign's inerrancy. Some assert that since the property and the property-bearer are already confirmed, it is the relation that is inferred. 12 8.1.0 Now on this matter "Some people claim that it is another property that is cognized through the sign's inerrancy." They claim that since from smoke we become aware of fIre that accompanies it, rather than becoming aware of a place possessing fire, it follows that we infer the fIre itself from the smoke. "Some assert that since the property and the property-bearer are already confirmed, it is the relation that is inferred." These people claim that since fire and its locus are well-known to people, the thing that is inferred from smoke is the relation between fIre and its locus.

9.0.0 If the sign is known to occur at the property, what else is inferred through it? If it's at the property-bearer, why is that not the thing that is inferred? 9.1.0 Let us fIrst answer the former view. "If the sign is known to occur at the property, what else is inferred through it?" If the inferential sign, smoke, is already known to be at the other property, fire, then what is the purpose of recalling the relation between smoke and fIre, and what is inferred through smoke? "If it's at the property-bearer, why is that not the thing that is inferred?" If fIre is inferred through the sensation of smoke in a locus that is connected with fIre, then why not say that the locus itself is

Chapter Six

Pramat;asamuccaya 2: J0

243

inferred to be possessed of fIre? For it is not the case that one is not aware of fIre there.

10.0.0 The two do not occur in the relation. We would hear the possessive case marker applied to the possessor. Inexpressible, it is known by implication. It has no relation with the evidence. 10.1.0 And to those who say that it is the relation that is the object of inference [we reply]: "The two do not occur in the relation." Fire and smoke do not occur in the relation, so this would amount to saying there is fIre whenever there is no occurrence of smoke. 10.1.1 Iinendrabuddhi (pllOa4) and Kitagawa Hidenori (1965:106-107) interpret this passage as follows. For an inference to be correct, the evidence smoke and the inferred property fire must reside in the same locus. Now when making an inference we do recall the relation of the evidence and the inferred property, but relata are the loci of a relation, not vice versa. Thus if the relation itself is regarded as the object of an inference, then the evidence, since it cannot reside in that object, would fail to have the first characteristic of legitimate evidence.

10.2.0 Furthermore, "We would hear the possessive case marker applied to the possessor." If the relation were the object of the inference, we would see the possessive case applied to [the word for] fIre, which has the relation: "The relation of fIre." [But in fact] we employ the nominative case: "Fire is here." A relation is "inexpressible" through its intrinsic properties; a relation is expressible only in terms of something else [namely, its relata]. That being the case, "it is known by implication." When we say "There is fire here," the relation [of fire to its locus] is expressed only implicitly. For the reasons stated above it is not what is inferable through smoke. "It has no relation with the evidence," since it is not the case that the [relation's] inerrant occurrence with smoke is shown elsewhere. Rather, [the inerrancy relation is shown to be of smoke] with fIre.

11.0.0 But it is the sign's inerrant occurrence with the property that is pointed out elsewhere. When known there it will make the property-bearer known to be endowed with the property. 11.1.0 Suppose it is argued that the above criticisms are invalid, because the locus endowed with fIre is regarded as what is inferable, despite the fact that smoke is not shown to be inerrantly related to that locus. So the same may be true of the relation as well. "But" this is not a parallel instance, because "it is the sign's inerrant occurrence with the property that is pointed out elsewhere. When known there it will make the property-

244

Pramar;asamuccaya 2: 11

On reasoning

bearer known to be endowed with the property." When one sees the inerrant relation of smoke with fire in one place, then by observing only smoke in a second place, it is possible to confirm that the second locus is possessed of fire on the grounds that wherever there is smoke, there is fire. Otherwise, we cannot account for it. [We cannot say for example] that a specific instance of smoke and a specific yet unconfirmed locus are inerrantly related elsewhere, but we can point out the inerrant relation generally, for what is indicated is that wherever there is smoke there is fire. Therefore, it is correct to say that although the inferential sign is shown to be in errantly related with the property to be confirmed, neither the property itself nor the relation is what is to be inferred.

6.4 On restricted and errant properties

12.0.0 Now we must consider the other property, and we must also explain the inferential sign. 13.0.0 An object has many properties. But we do not become aware of them all through the inferential sign. By a process of eliminating others, the sign makes known those with which there is a relation. 13.1.0 Concerning a property-bearer that displays a sign that is restricted to a property and is thereby confirmed to be in possession of that property: "An object has many properties. But we do not become aware of them all through the inferential sign. By a process of eliminating others, the sign makes known those with which there is a relation (rjes 'brei; *anubandha)." We cannot by means of the smoke become aware of what kind of special features the fire has, as for example what kind of flames it has or what temperature, because the sign is errant with respect to those features. But one does become aware of those things that are related, things without which no fire exists, such as the fact of being a substance (rdzas nyid; *dravyatva) and the fact of possessing qualities (yon tan nyid; *gur;atva). One is aware of these properties as incompatible with nonsubstance, etc. For example, the cognition of fire is in accordance with things that being related to it can only rule out what is not fire. One does

Chapter Six

Prami'ifJasamuccaya 2: 14

245

not observe them all in fIre's absence, but one does observe [some of them] in things other [than fire].

14.0.0 Possession of qualities, aroma, sweet fragrance and a particular sweet fragrance: taken in this order, each increases the notion of a lotus by ruling out such things as non-substances etc. 14.1.0 Possession of qualities rules out non-substances, and possession of aroma rules out non-earthen substances, and possession of a sweet fragrance rules out thin§s that stink, and a particular sweet fragrance rules out what is not a lotus. I Each of these eliminations makes [the lotus more clearly] known. Otherwise,

15.0.0 If the sign made the object known by a means of confirmation similar to direct sensation, then either the object would not be known at all, or it would be known in its entirety. 15.1.0 If a sign revealed [an object] at a time later in the same manner as the [earlier] sensation of smoke at fire, then it would not reveal it anywhere; the sign is as unobserved in every fIre as it is unobserved when there is no fire at all. 15.1.1 The point of this passage is to show that the inferential sign and the property inferred though it must be universals. The reasoning is explained by Jinendrabuddhi in his {lkii (p1l3b3 -114a3). The reasoning as spelled out there can be recapitulated as follows. At the time when we observe smoke and fIre together, we necessarily observed a particular instance of smoke (SI) with a particular instance of fire (PI)' That instance of fire (PI) is never seen with any other instance of smoke (Sn' where n is other than 1). Now suppose the property to be inferred were just that particular of fire (PI)' Since every subsequent instance of smoke (S2, S3, S4"'8o) is as absent from (PI) as they are absent from places where there is no fIre at all, we should never be able to infer the presence of the fIre (Pj) from the sign (Sn)' Therefore, if there is any inference at all, it can only be of a umversal fire-in-general. By a similar line of reasoning, it can be shown that the inferential sign can function only as a universal. For the particular instance of smoke (SI) is as absent from (P2, F3 , F4 ... Fn) as it is absent when there is no fire at all, and not being related to (Po, where n is other than 1) it can of course never be the grounds for inferring it.

15.2.0 And if the sign revealed the object in the same way that sensation does, then it would also reveal it as possessed of a specifIc flame and a specific temperature and so on. Since [the sign] makes [the inferred property] known generally, by ruling out non-fire, we know that by means

246

PramiifJ-asamuccaya 2:16

On reasoning

of the confirmed property [that is, the sign] we become aware of only this [general] form, but not the particular form.

16.0.0 But there really is no universal. Because we do not observe it throughout its substratum, nor do we see it outside its substratum. But if it is observed in each of its substrata, it is divided. 16.1.0 First of all, there exists no universal frrehood in addition to the fire. Even if it exists, it is impossible to observe it, because one cannot observe it in its entire substratum. We observe that no property that is common to many substrata, such as duality, can be cognized in its every substratum [that is, in every pair]. Some say a universal is due to resemblance, but there is also no resemblance. 14 16.1.1 There are some who argue that the notion of a universal is based upon a real resemblance that exists and is observable in particulars. This resemblance is supposed to be a feature of the objects as they are discovered by experienced, as opposed to being superimposed by the mind upon the objects experienced. According to this view, resemblance can be regarded as an entity that exists in addition to but not necessarily independent from the particulars that resemble one another.

16.2.0 Some say that if a universal is known in a single substratum it is known in all, but in this case it would be plural like the substrata themselves. 16.2.1 Each school of Indian philosophy had its own way of dealing with the puzzle of how a universal construed as a single, undivided and unchanging entity can reside in a plurality of changing entities. This topic comes up for discussion at greater length in the fifth chapter of the Pram[IlJ£lsamuccaya. But here Ditmaga confines himself to pointing out difficulties in accepting the view that universals are real entities that exist in addition to the particulars in which they are supposed to inhere. For more on how Dinnaga's school and other schools of Indian philosophy treated universals, see Dravid 1972.

17.0.0 Thus, although the sign has several properties, it is really only part of them, namely, those that are not errant from the thing indicated, that make it known. The others do not. 17.1.0 And similarly, only part [of the totality of properties of the sign] reveal the thing to be inferred from it. "Thus, although the sign has several properties, it is really only part of them, namely, those that are not errant from the thing indicated, that make it known. The others do' not." In the case of smoke, it malces fire known only through part of its properties,

Chapter Six

PramalJasamuccaya 2: 17

247

namely, those such as the fact of being smoke or the fact of having a smoky colour that do not occur except with fire; but not being a substance, because this property is errant. 17 .1.1 The fact of being a substance is not restricted to fires, since it occurs in all sorts of material things other than fire. This being so, when the property of being a substance occurs in a body of smoke, it plays no role at all in the inference of the presence of fire from observed smoke. But there are properties that belong to smoke and which never occur anywhere unless fire is also present. Such properties are the fact of being smoke, certain colours that occur only in bodies of smoke and whatever other sensible properties there may be that occur only in the presence of smoke and hence only in the presence of fires.

17.2.0 The following verses give the essence of the above topic.

18.0.0 The sign makes known also that which is related to the sign-bearer. It does not make the latter's particularities known, because the sign is errant from them. 19.0.0 An object related to the sign does not make the signbearer known, because it is errant from the latter. After we become aware of what is particular [in the sign] it makes one aware [of what is inferrable through the sign].

6.5 Non-symmetry of restriction and pervasion

20.0.0 Although the relation of the co-existing sign and signbearer is located in both of them, it occurs in the manner of a content to its container rather than in the manner of things ill contact. 15 20.1.0 Now one might get his idea: since the relation between a sign and a sign-bearer resides in both relata, just as physical contact (sarhyoga; sbyor) [resides in the two things contacting one another], it follows that the property that is a sign is interchangeable with the property that is a signbearer. But that is not the case. "Although the relation of the co-existing sign and sign-bearer is located in both of them, it occurs in the manner of a content to its container rather than in the manner of things in contact." Although the relation [between a container and its contents] is one that occurs in both relata, the container does not assume the role of the content,

248

Pramar;asamuccaya 2:20

On reasoning

nor does the content assume the role of the container. In just the 'same way, the sign does not in any case assume the role of the sign-bearer, nor does the sign-bearer ever assume the role of the sign. In the case of physical contact, on the other hand, the second relatum is just like the first. But such is not the case with this. Thus,

21.0.0 The sign-bearer must occur where the sign occurs, and the sign occurs only where the sign-bearer occurs. When this rule is reversed, there is no relation of sign and sign-bearer. 21.1.0 Since the sign-bearer must exist at the sign, it is possible by means of smoke to become aware of [fIre's] being a substance as well as its being a fire, but not to become aware of its temperature. Since the sign exists only at the sign-bearer but not elsewhere, such attributes of the smoke as its being smoke or its being smoke-coloured can make one aware [of fIre], but [smoke's] being a substance cannot. Thus when this rule is reversed, there is no relation of sign and sign-bearer.

22.0.0 Now if one claims that a sign, such as the fact of being homed, pervades the sign-bearer, then some of the sign occurs outside the sign-bearer. Because it is the pervader, it cannot make [the sign-bearer] known. 22.1.0 Now suppose the sign does in fact occur wherever the sign-bearer occurs, as for example the fact of being created which occurs wherever transitoriness occurs. "Now if one claims that a sign, such as the/act of

being horned, pervades the sign-bearer, then some of the sign occurs outside the sign-bearer. Because it is the pervader, it cannot make [the sign-bearerJ known." If only some of the sign occurs at the sign-bearer, then by virtue of the fact that it pervades the latter, it does not make the sign-bearer known. For example, although the fact of being horned pervades cows, it is not capable of making cows known. But the fact of being a cow, since it does not pervade [the fact of being horned] does make this latter fact known. Why?

23.0.0 Since what does not occur in the complement depends on the pervader, although a sign pervading the sign-bearer occurs, it is not evidence. 23.1.0 The fact of being created pervades objects that are transitory, but it cannot make them know except by excluding impermanence. Similarly, the fact of being transitory pervades objects that are created, but it cannot

Chapter Six

Pramiir;asamuccaya 2:'23

249

make them known except by excluding the fact of being uncreated. Therefore, since the fact of being created pervades transitory objects, the fact of being uncreated cannot occur in them. Therefore, transitoriness can make the fact of being created known by excluding the possibility of being uncreated, but it cannot make it k...'1own [through pervasion]. The following verses summarize the above points.

24.0.0 Since that which is perishable is pervaded by the condition of being created, what is perishable is not uncreated. It is not claimed that on the basis of this pervasion perishability is absent in produced objects. 25.0.0 That the condition of being homed pervades cows excludes the condition of being unhomed [from cows]. That cows are pervaded by the condition of being homed does not exclude non-cows [from the condition of being homed.] 25.1.0 This concludes the explanation of inference for oneself.

-- Notes--

Chapter 6. PramiilJasamuccaya ll: "On reasoning" 1. The quotation is from Vai.§e~ika Sutra 2.1.10: "Na ca dr~tlinaIh sparsa ity adr~taliIigo vayuJ:!. (And since [its] touch is not that of things that are seen, wind has a sign that is unseen.)" Candrananda (Kal)ada 1961 ed., p. 12) says: "Yadi khalv ayaIil k$ityadisparso 'bhavi$yad gandharasarupailJ sahopalabhemahi, na caivaIh, tasmat prthivyadivyatiriktasya vayor liIigam. (If in fact this [wind]) had the touch that earth and other [elements] have, then we would apprehend it with its smell, taste and visible properties. But it is not like that, and therefore wind, which is distinct from earth and so forth, has a sign [by which it is known]." See also Udayana's KirafJiivali on this passage in PraSastapada 1971 ed., pp.

56-57. 2. The phrase "confirmed by a process of elimination" is a translation of the Tibetan phrase "yongs su lhag pas grub pa" that appears in Kanakavarman's translation or "kun gyi rjes thogs" that appears in Vasudhararak$ita's. Kanakavarman's Tibetan phrase could tentatively be retranslated into Sanskrit as "parise~el)a (or parise$yat) siddha," but the rendering by Vasudhararak$ita is more resistant to understanding.

250

PramiilJasamuccaya 2:25

On reasoning

3. My translation of this passage in an earlier publication (Hayes 1980:249-250) was quite different from this. The basic issue of this passage was badly distorted in that translation because of my labouring under a fundamental misconception about the Vaise~ika system of thought. In that earlier translation I made the very serious blunder of reporting that the quality touch was said by the Vaise~ikas to inhere in only one substance, wind. But I have learned since then that the real issue here is that the Vaise~ikas taught that the touch that inheres in wind is distinct from the kind of touch that occurs generally in the other elements. A good overview of the Vaise~ika system, incidentally, is provided in both Matilall977 and Potter 1977 ed., pp. 212-220. 4. For the history of this idea in the literature of the Sanskrit grammarians see Brough 1951 and 1953. 5. Randle (1926: 17) cites the Sanskrit for the first line of this verse, and the second line is quoted by Uddyotakara in Gautama 1967 ed., p. 301. The whole verse reads: aptavakyavisamvadasamanyad anumanata anumane 'tha tattulye sadbhavo nastitasati. 6. Kitagawa (1965:93) follows Vasudhararaksita's syntax here, which places phyogs and gtso bo as two things whose natures are denied as objects of inference. Kitagawa translates "phyogs" as standing for Sanskrit "dis," meaning direction. But Iinendrabuddhi (p105a 8) supports Kanakavarman's translation, which makes, I think, much better sense. In this interpretation "phyogs" should be taken as being the translation of the original Sanskrit "pa/qa." in the sense of hypothesis. 7. For these fIrst sentences I have followed the Tibetan translation of Vasudhararaksita (D29a4), because the translation by Kanakavarman (pllla6 ) appears to be faulty here. It is not formed of grammatically complete sentences and seems to be lacking at least one crucial phrase. It reads: "rjes su dpag pa ni chos khyad par can gyi chos can yin tel de la dus phyis chos kyi spyi'i tshul gyis mngon sum mam Ijes su dpag pas mthong ba'ol Ide'i rigs la yang mtha' dag gam phyogs gcig yod pa nyid dol I (Inference (sic) is a propertybearer qualified by a property. Later one observes, either through sensation or inference, a property there in its general form. Presence wholly or partially also in that universaL)" 8. The material supplied in square brackets in the translation is taken from the commentary of Jinendrabuddhi (P108b3), who expands the verse to read "ji ltar rtags kyi yul can gyi shes pa," which could be retranslated into Sanskrit as "lingavisayam jilanam." 9. I have followed the translation ofVasudhararaksita for this clause. 10. The term "psychologism" is used in a variety of ways, but I am using it to refer to the issue of whether questions of logic, aesthetics and epistemology can or should be reduced to a study of the way that human beings in fact think or ought to think. Haack (1978:238-242) has a good discussion of this issue. Haack herself seems to favour a kind of what she calls a "nominalist version of weak psychologism," according to which logic is normative and prescribes how we should think and is therefore psychologistic but is a weaker form of psychologism than the view that logic describes how we in fact think. What is nominalistic about Haack's interpretation of weak psychologism is that she prefers to think of reasoning and believing as a complicated relationship between a person and sentences rather than as a relationship between a person and independently existing relations among propositions. Since the issue of psychologism has been important during

Chapter Six

Pramiil].asamuccaya 2:25

251

the past two centuries of European-based philosophy some discussion of this issue is bound to playa role in the comparison of Indian and European systems of logic and epistemology. I do not at this point feel quite prepared to go deeply into that issue, but perhaps it will be the subject of a future study. 11. The original Sanskrit for this verse has been discovered by Chatterji (1929-30): .!qtakatvad dhvanir nityo miirttatvad aprameya~ amiirtaSravaI).atvabhyam anityaS ciik$u~atva~. 12. The Sanskrit originals for the next four verses are quoted from Vacaspatimisra's Ttitparya{fkii in Gautama 1967 ed., p. 320. They are also cited and discussed by Vidyabhii~aJ:Ja 1921:281-282, Randle 1926:18 and Matilal 1968. Matilal also provides information on how the verses were discussed by Vacaspatimisra and later Naiyayika commentators, and on the basis of this information suggests alterations in Randle's tentative translations. The Sanskrit for these verses reads:

kecid dharmantararil meyarillingasyavyabhicara~ sambandharil kecid icchanti siddhatvad dharmadharmil)olJ lingam dharme prasiddharil cet kim anyat tena mIyate atha dharmil)i tasyaiva kimartha nanumeyata sambandhe 'pi dvayam nasti ~asthI sruyeta tadvati avacyo 'nugrhltatvan na casau lingasarilga~ lingasyavyabhicaras tu dharmel)anyatra disyate tatra prasiddharil tadyuktam dharmil)aril gamayi~yati. Some quotations of verse eleven read "drsyate" for "disyate," but the Tibetan translations all support the latter reading, using various forms of the root "ston," meaning to teach or to show. 13. In translating this passage I have followed the Tibetan of V asudhararak~ita rather than that of Kanakavarman. 14. This passage is rather obscure in both Kanakavarman and Vasudharar~ita, so I have had to rely entirely on linendrabuddhi (P114b 6). The passages under discussion read as follows. KI13b 1: "gnyis nyid la sogs pa du rna dangl thun mong ba mams kyi rten rna bzung bar 'dzin pa ni mthong ngol gang dag 'dra ba phyir smra ba'i 'dra ba'ang rna yin nol" V32a 4: "du rna mams las gnyis nyid la sogs pa'i thun mong ba ni yod pa rna yin nol gang dag spyi mthog zhing gzung pa po yang rten rna gzung pa po dang mtshungs shing 'dra bar 'gyurrol" 15. The Sanskrit original for this verse has been discovered by Katsura: sambandho yadyapi dvi~tal) sahabhiilingalinginolJ adharadheyavad vrttis tasya sarhyogivan na tu.

Chapter 7 PramiiIJasamuccaya V:

"On the nature of signs in language" 7.1 On the question of what verbal symbols make known

1.0.0 Verbal communication is no different from inference as a means of acquiring knowledge. For it names its object in a way similar to the property of having been produced, by precluding what is incompatible.! 1.1.0 We have discussed the two means of acquiring knowledge. But some claim that verbal communication is an additional means of acquiring knowledge. "Verbal communication is no different from inference as a means of acquiring knowledge. For it names its object in a way similar to [an inferential sign such as] the property of having been produced, [which indicates its object, namely, the property of being impermanent] by precluding what is incompatible [with the indicated property]." Like the property of having been produced, a linguistic sign reveals part of the object to which it is applied, namely the part with which it is necessarily related, and it reveals this part by excluding what is incompatible [with what is indicated by the verbal symbol]. Therefore, it is no different from inference. 1.1.1 At the outset of his discussion of the means of acquiring knowledge (pramiifJa),

Ditmaga says "The two means of acquiring knowledge are sensation and inference. There are only two means, since knowable objects are of two types. There is no type of knowable object besides the peculiar attribute and the general attribute. We shall show that sensation has the geculiar attribute as its object, while inference has as its object the general attribute." Diimaga argues repeatedly in the fIrst two chapters of PramiifJasamuccaya that sensation and inference are mutually exclusive, that is, that no knowable object can be known by more than one of these two means. In the present chapter he will pursue the point that sensation and inference are also

252

Chapter Seven

Prama/Jasamuccaya 5:1

253

exhaustive, that is, that there is no means of acquiring know ledge other than these two. 1.1.2 The orthodox Brahmanical schools of Indian philosophy were those that accepted the authority of the Brahmanical scriptures, which were regarded as capable of giving rise to knowledge of matters beyond the scope of ordinary reasoning or sense experience. Jinendrabuddhi cites as a representative of these schools the views of the putative founder of the Sari1khya school, Kapila, according to whom verbal testimony (siibda) is a means of acquiring knowledge of remote objects. 3 This is probably an allusion to the view expressed in, for example, Samkhya Karikti: "An object that is beyond the range of the senses that cannot be established even through that [inference] is established through reliable persons and scripture. ,,4 The author of the Yuktidfpika commentary to the S3.ri1khya Karika explains that it is through verbal testimony that one gains knowledge of such things as heaven (svarga), deliverance (apavarga) and divinity (devata).5 A position very similar in sentiment to the S3.ri1khya view is advanced by Bhartrhari, who says: "Nor does reasoning establish sacred duty independently of scriptures. Even the knowledge of the seers was preceded by scriptures.',6 A little further on he adds "The special powers of the ancestors and of the Rak~asas and of the Pisacas, which derive only from their deeds [in former lives], are well-established facts lying beyond the realms of both sense experience and inference.',7 Thus, says the Brahmanical philosophers, it is necessary to regard scriptures as a means of acquiring knowledge, distinct from sense experience and inference, in order to account for our knowledge of a whole range matters that cannot be known by ordinary means. In opposition to this view, Dmnaga has two replies. First, some things mentioned in scripture, such as heaven and primordial matter (pra/q"ti) , are utterly unknowable; belief in them is not knowledge at all, but only unjustified belief. Of these things it is inappropriate to speak of a means of acquiring knowledge (pramal)a). Second, some things mentioned in scripture are knowable, but our knowledge of them turns out to be justifiable by ordinary sense experience or inferences that rest in the final analysis on sense experiences. But in this case, he argues, there is no justification for regarding verbal testimony as the source of our knowledge, for the source of our knowledge is one of the two ordinary means of acquiring knowledge. It is interesting to note, incidentally, that DharmakIrti implicitly rejects Dinnaga's position by returning to a position very much like that of Bhartrhari and Kapila in that he argues that the Buddhist scriptures should be accepted as authoritative when they speak of such things as heavens, hells, the details of karman and other matters that are beyond the range of sense experience and reasoning; the basis for the trustworthiness of the Buddhist scriptures according to DharmakIrti is that they are proven to be trustworthy in all areas in which what they say can be confirmed by reason and experience. DharmakIrti' s stand on this matter is discussed in some detail in Steinkellner 1982, Hayes 1984 and Tillemans 1986. Tillemans also discusses the views of such Tibetan commentators as Tsang kha pa BIo bzangs grags pa and Rgyal tshab Dar rna riD chen and of the Chinese translator Xuanzang. 1.1.3 Kamalaslla quotes this verse and goes on to explain that in the Buddhist view the relation between verbal symbols and their objects is not regarded as different in kind from the relation between the evidence (hetu) invoked in the inferential process and what the evidence indicates. On the contrary, say the Buddhists, linguistic signs are but one type of evidence, and also what one learns through a sentence is of the same nature as what one learns through a piece of evidence in an inference. In each

254

PramalJasamuccaya 5:1

On linguistic signs

case what is learned is general in nature rather than particular, and therefore verbal communication is to be subsumed under inference. 8 So far, this is a fair account of Ditmaga's overall views on the matter. Where KamalaSIla deviates from Ditmaga's position is in his going on to say that the meaning of a sentence is understood from the sentence owing to the fact that there is a causal relation between the sentence and what its speaker intends to express by it? Describing inference in terms of causal relations is a feature of DharmakIrti's but not Dirinaga's theory of inference. Given the difference between Dirinaga and his later interpreters on this point, it may be worthwhile to review here what Dirinaga said about how evidence functions and then see how in his view a linguistic sign may be regarded as but one type of evidence. In the second chapter of PramalJasamuccaya, in verses 12 and 17-19, Dirinaga establishes the principle that presence of a given property PI in a certain locus L can be used as conclusive evidence of a second property P 2 only if PI is known to be restricted to the set of loci in which P 2 is present. In other words, the set of individuals that collectively constitute the extension of PI must be contained in (be a subset of) the set of individuals that constitute the extension of P2 . If this condition is met--that is, if the evidence (hetu) PI occurs in no individual that is in the counterextension (cf. vipak:ja) of the property to be confirmed (sadhyadharma) Pz--then the presence of the evidence in a given locus makes known that the locus is within the extension of the property to be confirmed. It makes this known by excluding (apohena) that locus from the counterextension (anya) of the property to be confIrmed. Similarly, when a linguistic sign is applied to a given Ob~.ect, the verbal symbol's application makes known some "property" (cf. svartha) of that object by way of excluding that object from the counterextension of the "p operty" that the expression names. Thus the function of a linguistic sign that is appli d to an object is said to be in all important respects exactly parallel to the functi n of a property used as evidence in an inference. The question of what sort of "pro erty" a linguistic sign names will be dealt with in some detail in the passages that follow. 1.1.4 Dirinaga says "a linguistic sign reveals part of the object to which it is applied. namely the part with which it is necessarily related, and it reveals this part by excluding what is incompatible." Any individual object to which an expression is applicable may be seen as a collocation of general properties, that is, as the locus of a number of different universals. As will be seen in greater detail below (verses 12 ff.), a linguistic sign does not make known every property of every object to which it is applicable. Rather, it makes known only that property or those properties that occur in every object to which the verbal symbol is applicable. In other words, a linguistic sign applied to an object tells about that object that it has a given property, a property that is never absent in any object to which the sign is applicable. Since it is our knowledge of the linguistic convention that a given expression is never applied to objects in which the property in question is absent, it is said that the expression makes the property known by excluding the objects to which the linguistic sign is applicable from the counterextension of the property. Or we may say what amounts to the same thing, that the linguistic sign makes a given property known in a given object by excluding from that object all properties that are incompatible with the property expressed by the expression. And so, as was explained above in 1.1.3. an expression functions in the same way as a property used as evidence in an inference to convey information about the objects to which it applies. An expression and an indicator property (that is, a property used as evidence) are similar in that neither makes its observer aware of any "unique aspects of an object or property-locus but rather only of aspects that the object has in common with many other objects. In Dirinaga's system, a cognition wherein the

Chapter Seven

PramlilJasamuccaya 5:1

255

cognizer is aware of a single property in the form of a sense datum is a case of sensation, whereas cognition of a multiplicity of loci bound together as a group in virtue of their possessing a property in common is a case of judgement or inference.

2.0.0 A general tenn does not express particulars, because they are unlimited in number, because it is errant, nor the relation nor the universal itself, because it is heard without a difference with words referring to particulars. 10 2.1.0 There are those who argue as follows. A general term expresses every one of its own particulars. But a particularizing expression is applied to what is so expressed in order to limit it. To those who hold such a view we reply "a general term does not express particulars, because they are unlimited in number." For, since the particulars are unlimited in number, it is not the case that each one can be associated with the expression; an expression that is not associated with an object cannot express that object, and so there is cognition of nothing but the expression's form. 2.1.1 "There are those who argue as follows. A general term Uatisabda; rigs kyi sgra) expresses everyone of its own particulars." But a particularizing expression (bhedaSabda; khyad par kyi sgra) is applied to what is so expressed in order to limit it." This view of the way that several components in a phrase function together is, according to Jinendrabuddhi, advanced by someone who rejects Ditmaga's claim that expressions indicate their objects through an exclusion or abstraction process . . According to this hypothesis under consideration, in a phrase consisting of two or more components in apposition, one component serves to pick out a relatively large class of objects by naming the generic property that just those objects have in common, while the other components serve to pick out specific subclasses or individuals within that larger class by naming more specific properties, that is, properties with smaller extensions. The English word "particular" has been chosen to translate the Sanskrit "bheda" (Tibetan "khyad par" or "tha dad"), which here signifies a part or division of a class and may be applied either to subclasses or to individuals, an individual simply being a class with only one member. Iinendrabuddhi (P271a2 ff") illustrates the view under discussion with the expressions "ma(:hara brahm~a" and "kauI;lQ.inya brahm~a," wherein the verbal symbol "brahmaI}a" picks out the class of people who have the property of being brahmans, while the particularizing expressions "mathara" and "kaul)Q.IDya" narrow the references to particular clans, that is, to subclasses of the brahman class that comprise just those individuals having the properties of being descended from Mathara and Kaul)Q.inya respectively. 2.1.2

"To those who hold such a view we reply 'a general term does not express particulars, because they are unlimited in number.'" Part of this verse was quoted by KamalaslIa in his explanation of the attempt by Santarak~ita to justify the claim that a particular (svalalqaQa, vyakti, bheda) cannot be made known through verbal symbols.! I Santarak~ita, who follows the essence of Dinnaga's argument very closely on this point, argues that verbal symbols do not make their listeners aware of a particular object, because it is not the case that any particular manages to be

256

PramalJasamuccaya 5:2

On linguistic signs

present on every occasion in which the verbal symbol's use is justified by linguistic convention and usage. An inferential sign, of which a verbal symbol is but one type according to Diimaga, reveals an object only if the object revealed by it is present on every occasion in which the sign is present; similarly, a verbal symbol expresses (vakti, pratipadayati) an object only if the object expressed (vacya, abhidheya) by it is present in the mind of the speaker each time the verbal symbol is correctly used, that is, whenever its use is justified (apta) by linguistic convention (sanketa). and usage (vyavahtira). To state the matter in the terminology of the logicians, there must be a pervasion (vyapti) of the occasions of the verbal symbol's correct usage (saliketavyavahtiraptaktila) by the thing expressed by the verbal symbol, just as in the inference of one property by means of another property there must be a pervasion of the indicator property by the property indicated. And so, if a verbal symbol functions as an indicator of some kind it can only indicate a class through some feature common to the plurality of individuals that belong to that class, even though the individuals are distinct from one another in their spatial and temporal locations, their actions, their potentials, and their various accidental properties. An individual does not pervade a class, and so an individual is not expressed--that is, made known or indicated--by the linguistic sign. This, according to Kamala.sIla, is the import of Santarak~ita's verses that read: A particular is not made known through verbal symbols, because [the particular] lacks pervasion of the occasions [in which the verbal symbols' use is] justified by convention and usage. Particulars do not include [that is, pervade] one another, because they are mutually distinct in time, place, activity, potentiality, appearance and so forth. Therefore, no object [such as a universal] that is known through a linguistic convention is found in practical life, nor is an object for which no linguistic convention is fixed [that is, a particular] known through an expression any more than a dissimilar object [to which the expression in question does not conventionally apply] is made known through that expressionP 2.1.3 "An expression that is not associated with an object cannot express that objec~nd so (upon hearing an expression that is not associated with its object] there is cognition of nothing but the expression's form." The expression's form refers to the configuration of sounds that constitute a particular expression. I3 In this passage and in verse three and its autocommentary, Ditmaga is following a series of ideas expressed in Bhartrhari's Vtikyapadfya in the section on relations in the third chapter.t 4 Bhartrhari states in Vakyapadfya 3.3.1 that three things are understood through spoken verbal symbols, namely, a) the cognition or idea that the speaker intends to convey, b) the external objects to which the spoken verbal symbols refer, and c) the forms of the verbal symbols themselves.t 5 What allows the hearer of the verbal symbols to cognize the speaker's intention and the external objects about which the speaker is thinking is a well-established relation between verbal symbol and referent; without this well-established relation, the hearer cognizes only the verbal symbol's forms. Ditmaga is arguing here that there can be no wellestablished relation between a term and each individual to which the term is applicable.

2.2.0 Moreover, [a general term cannot express particulars] "because it is errant" [with respect to any given particUlar]. Since the verbal symbol

Chapter Seven

Pramiil;lasamuccaya 5:2

257

"real" applies to qualities and so forth in the same way that it applies to substances, it does not explicitly express [either substance or quality], but rather its errancy gives rise to uncertainty [as to whether, in a given case, the verbal symbol is expressing a qUality or whether it is expressing a substance]. 2.2.1 In this passage we can clearly see the contrast between tw00types of relation that an expression can have to an object. The two relations in question are the vrtti relation and the viicaka relation. An expression has a vrtti (or pravrtti) relation with an object in case the expression is conventionally applicable to the object. But an expression is viicaka, expressive, of a given object only if the listener becomes aware of exactly that given object upon hearing that expression used. It is one of the features of any generic expression that one determines its applicability even. to individuals that one did not know to exist at the time when one first learned how to use the generic expression. From this it follows that although a word may be determined to be applicable to a given particular, it is not the case that the word expresses that particular, for one may hear the word used and understand it but still not be made aware of exactly that particular. It is this distinction between a word's applicability and a word's expressing that allows Diimaga to assert on the one hand that particulars are inexpressible and to maintain on the other hand that there is some relation between language and states of affairs in the world such that some statements are true--that is, correspond in some way to the facts--and others false. As has been explained above, it is DiI'lnaga's contention that a verbal symbol expresses just that to which it is restricted, just as an inferential sign (lii/ga) reveals or makes known only those properties to the loci of which the sign is restricted. A verbal symbol, such as a word, is restricted to that from which it has inerrancy (avyabhiciira). A word is errant from a given thing if it can be correctly used even when that thing is not present. But a general term is not restricted to any given particular, for it is applicable to other particulars that are incompatible with that given particular. To state the matter in Dirinaga's terminology, a general term (jiitisabda; rigs kyi sgra) has errancy (vyabhiciira; •khrul paY with respect to all particulars (bheda; khyad par) to which it has applicability (vrtti; 'jug paY and therefore a general term is not expressive (na viicaka; brjod par byed pa rna yin) of any particular. 16

2.3.0 Some think that a verbal symbol expresses either a) just a universal or b) just the [universal's] relation to its particular instantiations. [They maintain this view] on the grounds of ease [of detennining the expression's relation to either of these two expressible objects] and on the grounds of [the expression's] inerrancy. But neither of these two alternatives is acceptable. :'Nor [does a general term express] the relation or the universal itself, because it is heard without a difference with words referring to particulars." If it were the case [that a general term expressed a universal or a relation], [the word "real"] would not be used attributively with '?lords like "substance" that refer to particulars [that is, to subclasses of the class of real things] in expressions such as "real substance," "real quality," "real action" and so on. But in fact we do observe such expressions [in which a general term is used attributively or predicatively

258

Pramiivasamuccaya 5:2

On linguistic signs

with an expression of narrower generality]. Neither reality nor relation is a substance or a quality, but rather both are properties of a substance or of a quality. As it has been said [by Bhartrhari]: "Two words, one expressing a quality and the other the locus of that quality, as a rule have different casemarkings; it is established that two words expressing [the same] substance are in grammatical agreement ~that is, have the same case-markings and the same number and gender]. "I 2.3.1 Throughout this and the following passages Ditmaga uses the expressi' n "tadvat," which was translated into Tibetan as "de dang ldan pa." Parthasara· imisra and Uddyotakara both explain "tadvat" as meaning jatimat, that is, the p 'cular in its role as bearer of a given universal. IS I have rendered it by the word "in tantiation." Some translators prefer to render "tadvat" as "concrete universal," an xpression that is often used to translate the notion of universal favoured by Arist tie in his arguments against the ideal universals endorsed by Plato. 2.3.2

As Jinendrabuddhi explains, advocates of the hypothesis that a general term expresses either a universal or a relation, this relation being construed as the fact that a given particular is a concrete instance of a given universal, believe that these hypotheses avoid the pitfall of having an innumerable plurality of objects of reference, because both a universal and a relation are regarded as unitary objects of reference. If this unitary universal is present in a particular, then the general term is applicable to that particular; but if the universal is absent, then the general term is inapplicable. By observing this pattern, the user of language can easily discern that the gener:al term expresses the universal. The same can be said mutatis mutandis of the relation.

2.3.3 "Nor the relation nor the universal itself, because it is heard without a difference with words referring to particulars." This compact phrase from Ditmaga's kiirikii can be expanded to read: "Nor does the general term express the relation of the universal to its instantiation nor does it express the universal itself, because the general term is heard without a difference of grammatical case endings with words referring to particulars, which are things other than the universal." As the autocommentary explains, if the general term is construed as expressing a single entity, whether it be a universal or the relation described in the previous note, then that single entity must be regarded as numerically different from the entity expressed by a verbal symbol of narrower scope such as "substance." This presents a problem, for if the general term "real" expresses one entity A, and the word such as "substance" expresses another entity B, then the two words cannot be coreferential; that is, "real" cannot be construed as an attributive or predicative adjective modifying "substance." When words that are co-referential occur in a Sanskrit sentence, their co-referentiality is marked by such features as agreement of case, number and gender. Words that are not co-referential lack such agreement and so appear with different case markers and different indications of grammatical number and gender. Now it so happens that in Sanskrit we frequently encounter such phrases as "sad dravyam," meaning either "a real substance" or "the substance is real," in which "sat" agrees in number, gender and case with "dravyam" and is thus regarded as being used attributively or predicatively with ':dravyam." This linguistic observation is seen by Ditmaga as arguing against the view that the two words express numerically different entities, namely, reality and substance

Chapter Seven

Pramr11Jasamuccaya 5:3

259

respectively, for two words are in grammatical agreement only when they are coreferential, that is, applied to one and the same entity.

3.0.0 But on this matter it is said that a relation is expressible [only] through properties of its relata. One expresses it by thus making it an [intentional] object, but an object is connected with other things. 3.0.1

Vasudhararak~ita's translation of the last foot of this verse inserts a negative particle and so must be rendered" ... but an entity is not to be connected with other things." linendrabuddhi (P273a7) quotes this last foot of verse in full, supporting Kanakavarman's reading. He goes on (P273b 2) to explain that whatever is an intentional object (dngos po; *vastu) is necessarily connected with some other thing. An intentional object is whatever takes on the status of an object by virtue of being either the subject matter of a cognitive act or a topic of discourse. Now when one hears it said that there is a relation, one immediately expects to hear exactly what is related to what. Because of this inevitable expectancy for further information that arises when it is stated that there is a relation, the reference to the relation itself is an incomplete reference. It is, says Iinendrabuddhi (P273a5) , as when one hears a conjunction such as "Devadatta and" and immediately expects to hear what it is with which Devadatta is conjoined. If this expectation is not fulfilled, then the reference to a conjunction is incomplete.

3.1.0 A relation is that which relates. Like desire and so forth it relates [a thing] to something else. Therefore, since a relation is expressible through the properties of its relata, there is no word expressing it through its own intrinsic properties. Therefore, it cannot be expressed by a general term either. 3.1.1 "A relation is that which relates. Like desire and so forth it relates [a thing] to something else." Vasudhararak~ita's translation of the opening sentence of the autocommentary differs considerably from Kanakavarman's. Iinendrabuddhi (P273b 1) quotes the opening sentence in a form that differs from both Vasudhararak~ita's and Kanakavarman's translations, but he suggests that the purpose of Diilnaga's remark here is to describe the essential nature of relationships in general. My translation, rather than being a literal rendering of Kanakavarman, is an attempt to capture the sense of linendrabuddhi's explanation. 3.1.2 'Therefore, since a relation is expressible through the properties of its relata, there is no word expressing it through its own intrinsic properties. Therefore, it cannot be expressed by a general term either." Here, as in his autocommentary to the preceding verse, Diilnaga is paraphrasing an idea expressed in Bhartrhari's Vakyapadrya. The passage in question is the opening set of verses of the section on relations (sambandhasamuddesa) in the third chapter of that work. Since several ideas important to the understanding of Diilnaga's semantic theory appear in that opening set of verses, they bear quoting in full here.

260

Pramaf)asamuccaya 5:3

On linguistic signs

A spoken expression makes known 1) the idea of the speaker, 2) an external object and 3) the form of the expression itself. The relation among these things is fixed. With reference to both the idea of the speaker and the external object, there is [on the part of the hearer of the expression] a clear understanding on some occassions and uncertainty on others. Error with respect to the perceived form of the expression itself does not occur. The connection between expression and meaning is made known through the genitive. [We say, for example] "this is the expression of that" and "this is the meaning of that." The [inseparable] nature of expression and meaning is also shown by that [genitive]. There is no designator that expresses a relation by the relation's own property. Because it is invariably dependent [upon its relata], a relation's own essence cannot be [directly] communicated. 19 In this translation of Bhartrhari' s verses and in this discussion of his ideas that follows, I have relied upon Helaraja's commentary to the VakyapadJya. 20 Three of Bhartrhari's doctrines that are germane to our understanding of Diimaga's theories are a) that the relation between expression and expressed is fixed, b) that the principal information that an expression conveys is of its own form, and c) that a relation is essentially inexpressible. Concerning the first of these points, Bhartrhari claims that the relation between words and the things symbolized by them is permanent and fixed in the sense that human beings neither create nor alter this relation. This view is rejected by Diimaga and the Buddhist epistemologists who follow him. Concerning the second point, Bhartrhari claims that the thing that an expression makes known most directly is just its own form, that is, the symbol class to which it belongs, for the hearer of the expression is most sure to be correct in his understanding of that. Understanding correctly what the symbol stands for carries with it a greater risk of error than hearing or seeing the symbol accurately. Concerning what it is that the listener is then able to infer upon hearing an expression corectly, there are two commonly argued views. According to some, such as Diimaga's interpreter DharmakIrti, the only thing that the hearer can infer with any certainty is the speaker's intention, the basis of the inference being the causal relation between the speaker's wish to say something and his saying it. Since, however, the speaker may be incorrect in the thinking that caused him to speak or may be deliberately misleading his listeners, there is no reliable inference to be made from his speech to actual states of affairs. DharmakIrti expresses this view in PramalJavarttika 1:4:

Language is a source of knowledge concerning the object that, being the content of the speaker's intention, appears in his mind. But lanrage is not necessarily connected to a real state of affairs. 2 Ditmaga says nothing that explicitly agrees or disagrees with DharmakIrti on this matter. According to other philosophers, however, language can, by virtue of a special expressor relation (vacyavacakasambandha) between language and states of affairs, serve as a basis of reliable inference of realities external to cognition. But whether one accepts that language serves as the basis of inferring only the speaker's

Chapter Seven

Pram[u;asamuccaya 5:3

261

ideas or that it serves as the basis of inferring external reality, the point remains that for Bhartrhari the foremost information that an expression conveys is of its own form, while all information about that for which the expression stands as a symbol is secondary or derivative. Concerning the third point, Bhartrhari claims that no relation is expressible by an expression that names a property that belongs uniquely to the relation itself. Evidently, what led Bhartrhari to his claim that relations are inexpressible was his observation of the fact that in Sanskrit each relatum in a relation can be expressed by its own lexeme, but there is no lexeme for the fact that the relata are related, this fact being expressed by an inflectional suffix that yields only one of the many forms of the lexeme to which it is affIxed. In the Sanskrit "suhrdas sukham," for example, there are two lexemes, SUH~D and SUKHA, naming the relata in a relation, but there is no lex erne naming the relation itself, this latter being indicated only by the suffIx -as that generates one of the forms of the lexeme SUH~D. A similar observation can be made of the English expression "the friend's happiness."22 In any case, Difmaga endorses Bhartrhari's claim that relations are inexpressible. And from that he draws the obvious conclusion that it cannot be the case that a general term such as "real" in "real pot" expresses the relationship between the universal reality and the particular pot that possesses that universal.

4.0.0 Not an instantiation; because it is grammatically subordinate; because it is applied figuratively; because it is not the case that there exists [a resemblance], since there is a difference in the form of the idea, as for example in metonymical application of "king" to a servant. 4.1.0 Some say it is just the instantiation of the universal that is expressed by a general term. [They hold this view] on the grounds of the possibility of co-referentiality [on the part of the term that expresses an instantiation] with a particularizing expression, and on the grounds of the ease of [determining the] relation [between such a term and its meaning], and on the grounds of [the expression's] inerrancy. 4.1.1

In the view under consideration, a general term is to be understood not as expressing each individual object that possesses a universal property, but rather the fact, common to a plurality of individuals, of possessing the same universal. This fact of possessing the same universal is itself a single referent. Thus it is a better candidate than the individuals for being that which a general term expresses, for it avoids the problem of innumerable referents. Furthermore, a general term is used only when this fact of possessing the universal is present Thus there is no question of its now applying to some referent a, now to some referent b where b is not identical to a. And [mally, according to this view, there is no problem with expression such as "real jug" in which the two words "real" and "jug" are coreferential. Two words can be co-referential only if there is a sense in which the two words, each of which has a different warrant for application (pravrltinimitta), nevertheless apply to one and the same object Now in the case of the expression "real jug," the word "jug" is applicable only to an object x such that x possesses the

262

Pramat;asamuccaya 5:4

On linguistic signs

jug-universal or jughood. But, according to the taxonomy of the Nyaya and Vaisesika schools, no object possesses jughood without also possessing reality, because jugs are a species of the genus reality. To put it another way, the word "real" is applicable to any object to which the word "jug" is applicable. So, to summarize the view under discussion, the phrase "real jug" ("san gha{alJ") can be understood as referring to any object x such that x possesses jughood and x possesses reality (sattvavtin ca gha!avtin ca arthalJ). This whole view is summarized neatly by linendrabuddhi (P273b7 ff.) in his subcommentary to this passage: Everything that has substantiality, which is the warrant for applying a word for a substance, has reality, the warrant for applying the word "real," and hence can have the word "real" applied to it; on that basis co-referentiality (gzhi mthun pa = stimtinadhikarwJya) is possible. If [general terms] express instantiations of the universals, one can say "real substance." It is not impossible to make a connection, because the connection consists in non-difference. On this theory [the general term] is not expressing particulars but rather the mere possession of the universal that is a common trait of those particulars. And there is no errancy [of the general term from what it is supposed to express], because when that possession of the universal is missing in a thing, then one cannot apply the [general] term to it.

4.2.0 To this view we reply: "[a general term does] not [express] an instantiation; because it [scilicet the word that expresses an instantiation] is grammatically subordinate.,,23 And if this theory is true, then the word "real" does not express a substance directly, but rather it expresses a substance to which the word's form [that is, REAL] and the universal [reality] are subordinate. Because it does not encompass such species [of reality] as the jug and so forth, there is no genus-species relation, in which case the word "real" and "jug" are not co-referential. If there is no encompassment [of the designation of one word by another word], there is no co-referentiality [between the words]. For example, since the word "white" expresses a substance only insofar as that substance is qualified by the quality white that is expressible by the word, the word "white" does not encompass such properties as sweet flavour, although such properties may also be in the substance that has the property white. Therefore sweet flavour is not a species of white colour. The same principle applies also in the case under consideration. 4.2.1

"It expresses a substance to which the word's form and the universal are subordinate." Grammatical subordination (upasarjana) refers to the linguistic phenomenon whereby a word, by virtue of its undergoing anyone of a number of morphological <;hanges and being positioned in a phrase in such a way that it qualifies another expression, gives up its own primary reference an.d refers instead to the object named by the word that the subordinate word qualifies, this qualified word being called the principal or head word (pradhtina). A standard example,

Chapter Seven

PramalJasamuccaya 5:4

263

given in JhalakIkar (1874) under upasarjana, is "dhanavan devadattaJ;" wherein the word "dh"ana" has acquired the secondary (taddhita) suffix MATup designating possession; by acquiring this possessive suffix and being juxtaposed with the other word "devadatta," the word "dhana" loses its primary significance of wealth and comes to refer instead to Devadatta, who is thus characterized as wealthy. Appealing to this grammatical consideration, Difmaga points out that if we construe the word "sat" as referring not to the universal satta (reality) but rather to any sattavat artha (reality-possessing thing) that instantiates the universal satta, then we must be prepared to acknowledge that "sat" has given up its primary signification and become grammatically subordinate to some other head-word. On first consideration, this account may seem unexceptionable, for we readily assent to the fact that in an expression such as "san gha(aJ;," the word "san" is subordinate to the word "gha{aJ;". But now we must ask what "gha!a" designates. According to the view under consideration, that general terms express instantiations of universals, the word "gha(a" must also be construed as referring not to ghalatva (jughood) but rather to any gha{atvavat artha (jughood-possessing thing); thus both "sat" and "gha{a," construed as referring to instantiations of their respective universals, become grammatically subordinate or upasarjana. The complications arising out of such a construal will be discussed in the immediately following section. 4.2.2 "The word "real" does not express a substance directly, but rather it expresses a substance to which the word's form and the universal are subordinate. Because it does not encompass such species as the jug and so forth, there is no genus-~li.ecies relation, in which case the word "real" and ''jug'' are not co-referential." 4 An account must be given of what it means in this context to say that one word encompasses (iikoipati) another, or that there is an encompassment (akoepa; 'phen paY relation between two words. It has already been pointed out (see 1.1.3 above) that the relation between words and the objects to which they are applied is seen as parallel to the relation between properties and the objects in which they occur. A further instance of that parallelism occurs in this passage of Difmaga and in several discussions of it by later Indian commentators. With respect to properties, there exists a pervasion (vyapti; khyab paY relation between one property P j and a second property P 2 in case PI occurs in every object in which P 2 occurs, or In other words in case the class of objects in which P2 occurs is a species of the genus PI' A parallel relation can be seen between words, this relation being called encompassment (akoepa). An akoepa relation exists between a word W I and a second word W 2 in case WI applies to every object to which W 2 applies, or in other words in case WI denotes a genus of which the class of objects denoted by W 2 is a species. That encompassment between words is seen as isomorphic with pervasion between properties is demonstrated by the following verse in Santarak~ita: Now it is claimed [by Dmnaga] that what is expressed [by a general term] is a thing possessing exclusion. In that case too it is hard to see the pervasion [of one expression] by another, since [the putatively pervading expression] is subordinate.25 KamalaSl:la glosses the word "vyapti" in this verse with the word "akoepa," thereby showing the virtual synonymity of the two terms "pervasion" and "encompassment." Bhattaputra Jayamisra's lucid commentary on this passage is worth quoting in full:

264

PramalJasamuccaya 5:4

On linguistic signs

The monk [Dillnaga] argues in the following wayan this matter; a general term such as "real" that designates only a thing that is qualified by the universal reality does not encompass the particular such as the pot that is in fact associated with that reality. For in making reality its warrant for application, the word "real" [construed in the above way] applies to a thing as a subordinate expression, not as an independent one. It expresses a thing qualified by all its qualifiers. And therefore, since the word "real" does not encompass a jug and so forth, it is not co-referential with the word expressing that jug. Because when a thing is designated as qualified by one attribute, the expression for that attribute is not co-referential with the expression for another attribute. For example, since the word "sweet," which designates the substance sugar qualified by sweetness, does not encompass other attributes associated with sugar such as the colour white, it is not coreferential with that expression for the colour white. That is, we do not find such expressions as "the sweet colour white." By the same token, the expression "real jug" would be untenable.26 linendrabuddhi (P274b5-275a2) makes the same formal point as layamisra's discussion, but the former offers a somewhat different interpretation of the illustrations. For the sake of comparison, here is what linendrabuddhi says: If the word "real" does not encompass such things as jugs, then they are not a species of it [that is, jugs are not a species of reality], because one does not encounter non-encompassed species. And because there is no genus-species relation, the words are not co-referential ....For example, since the word "colour" does not encompass sweet taste and the like, the latter is not a species [of colour], and consequently there is no co-referentiality between the expressions for them; we do not say "the colour sour." But if there is pervasion [of one property by the other], then there is coreferentiality [between the expressions for those properties]; we do say "the colour blue" ....The quality expressed by the word "white" is the colour white, for the colour white is the warrant for using the word. Since the word expresses a substance only insofar as it is qualified by that colour, it does not encompass the sweetness in that very same. substance, despite the fact that sweetness does occur in it For the sweetness, unlike whiteness, is not a warrant for applying the word "white." Uddyotakara (1916 ed., p. 320) and Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 598) also give summaries of DiIinaga's argument on this point, but neither of these summaries is as thorough as the two quoted above.

4.3.0 Furthennore, [a general tenn cannot express only an instantiation of the universal], "because it is applied figuratively" [to the instantiation]. The word "real" literally expresses either its own word-fonn or a universal; in being applied to one of those two things, it applies to an "instantiation

Chapter Seven

Pramaf)asamuccaya 5:4

265

metonyrnically.27 Any object to which an expression is apt'lied metonymic ally is not the thing literally expressed by that expression. 4.3.1

This passage is discussed by Uddyotakara and quoted verbatim by Jayamisra. Uddyotakara's paraphrase is very similar to the original sentence, except that he makes no reference to the theory that the word "real" refers primarily to its own word form. His paraphrase goes "The word 'real' applies primarily to reality; in being applied to that, it is applied metonymically to the- instantiation [ofrealityJ. ,,29 Jinendrabuddhi (P27Sb 4) refers to the· theory that a general term primarily expresses its own word-form, and consequently even reference to the universal is metonymic. What this means is that any given utterance of a word, that is, a word token, expresses fust the ideal word form or lexeme of which it is an instance. See p. 302 for an account of lexemes. In any case, whether one holds that the thing primarily expressed by a general term is the word form or the universal for which the word form is a symbol, the further application to an instantiation of that universal is regarded as metonymic, and thus the instantiation is a derivative or extended meaning.

4.4.0 [But even the hypothesis that a general term applies metonymic ally to an instantiation is untenable], "because it is not the case that there exists" a resemblance. Qualitative resemblance, whether it arise from the transfer of a notion of from the influence of the quality, is absent from the instantiation. 3o 4.4.1 In the passage 4.3.0 Diimaga argued that even if we grant that a general term applies metonymically to an instantiation, it cannot be accepted that the general term expresses the instantiation. But in 4.4.0, according to Iinendrabuddhi (P276a1), Diimaga undertakes to show that we should not even go so far as to grant that metonymic application to the instantiation occurs in the first place. His strategy is to demonstrate that figurative application of a word W that literally expresses one object a to another object b is possible only if a and b have some similarity or resemblance (sarupya; 'dra ba). This of course leads to a discussion of what conditions are necessary in order to enable us to say that two things resemble each other. According to Diimaga, two things can be said to resemble each other if they either actually have a similar quality or if they each seem to have a similar quality. If two or more things actually have a similar quality, then we can be conscious of each separate occurrence of that qualitative resemblance in a sequence of individual loci of that qUality. '!pat is, we can say "This rose is red, and this scarf is red, and this teapot is red." This transfer of a notion from one thing to another is discussed in passages 4.5.0 and 5.1.0 below. The other possibility, that two things only appear to have a similar quality, is taken up for discussion in 5.1.0 and 6.1.0 below. The stock example of two things appearing to have similar qualities is the notion that arises with reference to a colourless crystal that the crystal is coloured, a notion that arises when the crystal reflects the colour of an object placed near it. About these two modes of qualitative resemblance more will be said in subsequent sections.

4.5.0 Why does it Esc. qualitative resemblance] not arise from the transfer of a notion? "Since" when there is a figurative application of a word, "there is a difference in the form of the idea [from the idea produced by its

266

Pramtlljasamuccaya 5:4

On linguistic signs

literal use}, as for example in metonymical application of 'king' to a servant.'<31 For example, when [the king] utters the word "self" with reference to a servant, saying "That man is my very self," it is not the case that the same idea arises with respect to the king and with respect to the servant. 4.5.1 This passage in Diimaga is very terse. Kanakavarman and Vasudhararak~ita differ in their treatment of this passage. First of all; we have in Kanakavarman the sentence "rag gang yin pa de ni bran yin no," which may be rendered "He who is subservient is a servant." This is followed by the observation that the idea differs when this statement is applied to a king and a servant. In other words, a king may be called a servant in virtue of his sharing with an actual servant the quality of servility, but when the word "servant" is thus applied to a man who is in fact a king, our notion of the sense in which the word is being used differs from the sense we get when the word is applied to a man who really is a servant This seems to be the gist of what Kanakavarman offers in his translation. But Vasudhararak~ita translates the sentence "gang mi de ni nga rang ngo," which comes out "He who is a man is I myself." It could very well be that the Tibetan word "mi" stands for the Sanskrit word "purw;a" in the sense of ordinary man or king's subject. It is difficult to be certain what example is being alluded to here, but it is tempting to think that the allusion is to an example used in the Samkhya system to account for how the word "self (atman)," which is used properly to refer to the unchanging, eternal and impassible essential man (puru:;a), can be applied metonymically to refer to the transitory and suffering complex of body and mind made up of evolutes of nature (pralqti) or, as it may also be called, the primordial matter (pradhiina). An example is cited from ordinary life, in which a king might say of his servant, who in fact does all the actual physical work that leads to the fulfillment of the king's wishes, "That servant is my very self. ,,32 The relevance of this whole discussion to the issue at hand, whether the word "real" in "real jug" can be understood as applying literally to reality and metonymically to the reality-possessing jug, is that we are not conscious of the word's being used figuratively on the basis of some qualitative resemblance between the word's primary referent, reality, and its putative secondary referent, the jug.

5.0.0 Because we do not speak sequentially as the white colour of a jasmine flower and a conch shell and so forth. If the universal occurred in the substance through the influence of a quality, the idea of the substance would be vivid without a notion of the quality's actual substratum.33 5.1.0 Even if the general term is applied figuratively to the instantiation of the universals, [it is not on the basis of qualitative resemblance] "because we do not speak [in this case of a single quality residing in its various loci) sequentially as [we do when we speak of] the white colour of a jasmine flower and a conch shell and so forth." We observe that we speak in sequence of those things of which we have similar idea; for example, we say "The jasmine flower and the water lily and the conch shell are white."

Chapter Seven

Pramar;asamuccaya 5:5

267

Since the word ["real"] is applied simultaneously to both the universal [reality] and its instantiation, there is no qualitative resemblance based on the transfer of an idea from one thing to another. "If the universal occurred in the substance through the influence of a quality, the idea of the substance would be vivid without a notion of the quality's· actual substratum." Suppose that a quality's likeness appeared in the particular instantiation through the influence of a quality of a proximate entity. If that were so, then the idea of a substance would be distinct independently of a vivid idea of its qualities. For the occurrence of the idea of colouring of a crystal does not depend on an understanding of the colour's actual substratum, because to the observer the ideas of the crystal's colouring and the colour of what is placed near it are not different. 5.1.1 Bhattaputra J ayamisra summarizes this argument as follows: The monk has spoken as follows on this matter: "One thing that occurs in another thing does so depending on assistance. If a universal resides in a substance by the same means as colouring occurs in a crystal, that is, through assistance consisting in the process of reflecting a quality (anuranjana), the assistance consisting in the process must be given to the substance. And then a notion would arise of just the substance that is assisted by the universal, but without any notion of the universal itself. For when a crystal coloured by a colouring is cognized, its cognition does not depend on anything else. But that is not so in the case under consideration. Therefore that which gives assistance does not occur in the substance. 34

6.0.0 It would follow that the cognition of all things would be false, because their characteristics would be mixed together. In the case of a plurality of universals and simultaneous cognitions, 7.0.0 influence is ruled out. If there were influence by the totality, then it would be observed as mixed together. The same criticisms apply also to the relation of particulars with the universal. 8.0.0 lbey too have the nature of an instantiation belonging to a universal. Hence it is untenable. But an instantiation is said to be nothing but a particular, and that has already been rejected. 35 6.1.0 Furthermore, "It would follow that the cognition of all things would be false, because their characteristics would be mixed together." All

268

Pram[iI;asamuccaya 5:6

On linguistic signs

verbal cognitions would be come inaccurate, like the cognition of a crystal, because they would be perverted by the characteristics being commingled in the object. Moreover, "in the case of a plurality of universals and simultaneous cognitions influence is ruled out." 7.1.0 When there is a manifold cognition of a thing as a possessor of qualities and of the qualities it has, such as white and so forth--for example, of a jug's being earth~n, a substance, real, white, red, odorous and so on--then influence by a quality is ruled out. Since there are no internal distinctions, the whole cannot be determined through a given aspect having the nature of a quality.36 Nor can one even experience the natures of the qualities as parts, because what in fact one is aware of is the nature of the jug as a whole. 7.2.0 "If there were influence by the totality, then it would be observed as mixed together." If there were simultaneous influence of everything, the jug and so on, on the totality, then since there would be no awareness of the individual natures of the jug and so forth, and since all natures are observed as one, the awareness of all elements would be confused. 7.2.1 In his elucidation of this passage, linendrabuddhi (P279b7 -280a8) outlines three hypothetical views on how the qualities of one substratum might be said to influence a proximate substratum if we take seriously the analogy of the crystal reflecting the qualities of an adjacent object. In discussing this point let me call the substratum to which the qualities actually belong the home substratum. And the substratum that picks up the qualities of a proximate home substratum I shall call the borrowing substratum. First of all, says linendrabuddhi, it might be said that just one of the many qualities of the home substratum is reflected in the borrowing substratum. Second, each of the individual qualities of the home substratum might appear as individual qualities of the borrowing substratum. And third, the entire package of the home substratum's qualities as a whole might appear collectively in the borrowing substratum. Having given these three possibilities, linendrabuddhi then represents Diirnaga as ruling out the fIrst possibility on the grounds that the qualities of a given substratum are not mutually distinct or separated from one another. In other words, we cannot isolate just one quality from the total impression we have of any object and say that that quality alone is reflected in the borrowing substratum. The second hypothesis is also ruled out on the grounds that we cannot be aware of the different qualities in isolation from one another, and hence cannot say that the individually perceived qualities of a home substratum are reflected as individually perceived qualities of a borrowing substratum. With these first two possibilities ruled out, the only remaining possibility is that the entirety of the home substratum's qualities appear as the totality of the borrowing substratum's qualities. But in this case the home substratum would be indistinguishable from the borrowing substratum. This third possibility and its repudiation are outlined in 7.2.0. That, then, is linendrabuddhi's understanding of the import of the passage immediately under discussion. The purpose of the whole passage beginning "with 5.1.0 and ending with 7.2.0 is to repudiate the contention that a general term might apply figuratively to an instantiation on the basis of qualitative "resemblance" that arises

Chapter Seven

PramiilJasamuccaya 5:7

269

out of a quality's actually occurring in a universal and only apparently occurring in the instantiation through reflection or some other such function of the proximity of the universal and itS instantiation.

8.1.0 "They too have the nature of an instantiation belonging to a universal. Hence it is untenable." Although it is expressed by a separate word fonn, a particular is necessarily expressed by a word having the nature of a universal, and therefore such a word applies literally to the universal and figuratively to the particulars. Hence, all the faults mentioned with respect to the instantiation can be mentioned in connection with this view. According to this view, given that even when it is the universal that is to be named, a substance is expressed by secondary attribution of the word's fonn to it, the instantiation can be expressed as real and so on by means of the fonn of the universal. 8.1.1 In the passage beginning with 4.3.0 and ending with 7.2.0, Diimaga has examined the possibility of applying a general term metonymically or secondarily to an instantiation of a universal. The purpose of the passage 8.1.0 is, according to Iinendrabuddhi (P280a8 ff), to show that all of the arguments adduced with regard to the figurative application of a general term, whose primary reference is to a universal, to an instantiation of a universal can also be adduced with regard to figurative application of a general term to a particular. In the former passage the question was whether the word "real" in "real jug" could refer primarily to the universal reality and secondarily to the instantiation thereof, namely, the realitypossessing jug. To this question Diimaga gives a negative reply; a particular can only be named by a generic term, and this generic term must refer primarily to the genus, say jughood, and only secondarily to the individual jugs. And this being the case, then all the aforementioned problems associated with secondary reference arise again. And finally, there are those who maintain that the word "jug" really expresses only the lexeme JUG directly, whereas the reference to the universal jughood is a secondary one. This view would also be subject to the aforementioned problems of secondary reference.

8.2.0 "But an instantiation is said to be nothing but a particular, and that has already been rejected." The repudiation which occurred above was our statement that a general tenn does not express particulars.37 And that statement holds good for the instantiation of a universal as well.

9.0.0 It is considered even with respect to the bare fact of possession that it is either a relation betwen the universal and its instantiation or that it is reality. Suppose the meaning of a

270

Pramar;asamuccaya 5:9

On linguistic signs

general term is an instantiation such as a jar that does 'not occur in other things such as cloth. 9.1.0 Suppose we say that a general term does not express a particular but rather that it expresses nothing more than the fact of possessing a universal? In that case, "It is considered even with respect to the bare fact of possession that it is either a relation between the universal and its instantiation or that it is [itself a universal such as} reality." The expression "bare fact of possession (tadvanmatra)" means "being a possessor of a given thing (tadvattva)," wherein the abstraction suffix TV A refers either to a relation or to an attribute. This is in accordance with the dictum: "[The abstract nominal suffixes], when suffixed to compounds, to words formed with primary derivational suffixes and to words formed with secondary derivational suffixes, name relations, unless [words of the aforementioned types are] idiomatic, or of the same phonetic form [as a word expressing a general property or a universal], or naming an' invariable relation."38 But concerning this we have already said "Nor [does a general term express] the relation or the universal itself, because it is heard without a difference with words referring to particulars." (See 2.3.0.) 9,1.1 "[t is considered even with respect to the bare fact of possession that it is either a relation between the universal and its instantiation or that it is [itself a universal such as] reality." This verse is quoted in both commentaries to Kumarila Bhatta's Mfmarhsaslokavarttika. Parthasarathimisra quotes it in his comment on Slokavarttika 131, while Bhattaputra Jayamisra quotes it after Slokavarttika 128. Jayamisra gives the following explanation: The monk, anticipating a view saying that the thing expressed may be just the state of possessing a universal, poses the following alternatives: is this state of possessing a universal a) the relation between the universal and its instantiation, or b) does it have the nature of a generic property? And after posing these alternatives he says that one can apply the previously stated criticisms of the views that a universal is expressed or that a relation is expressed. But what he says here, namely, "it is considered .even with respect to the bare fact of possession that it is either a relation or reality," is equally applicable to his own hypothesis ~that a word expresses the bare fact of possessing exclusion].3 9.1.2 "[The abstract nominal suffixes], when suffixed to compounds, to words formed with primary derivational suffixes and to words formed with secondary derivational suffixes, name relations, unless [words of the aforementioned types are] idiomatic, or of the same phonetic form [as a word expressing a general property or a universal], or naming an invariable relation." This dictum probablYo originated with Bhartrhari and is quoted several times in subsequent grammatical literature.40 linendrabuddhi (p281a 8-281b5) offers a lengthy account of the meaning of the

Chapter Seven

Pramar;.asamuccaya 5:9

271

dictum, which I shall paraphrase here. According to the grammatical tradition, the abstraction suffixes (bhiivapratyaya; dngos po' i rkyen) TV A and TAl, name either a universal (jtiti) ot a relation (sambandha). When affIxed to a nominal compound (samasa; tshig sdud) such as "raja-puT/qa," says Jinendrabuddhi, the result is the expression "rajapuru$atva (rgyal po'i skyes bu nyid)," which designates a relationship between a master and his servant (svasvamibhava; rang dang bdag po'i 'brei pay. When the abstraction suffix is added to a word that has been formed from a verb root plus primary suffix (kr'tpratyaya) such as J.lvuI, the result is a word that names the relation between an action and some factor in the action (kriyakarakasambandha; bya ba dang byed pa can gyi 'brei paY. For example "pacakatva (,tshed par byed pa nyid)" (PAC + J.lvul [--> -AKA] + TVA) designates the relation between the action (kriya) of cooking and the cook who is the agent (kartr) of that action. When the abstraction suffIx is added to a word that has been formed by adding a secondary suffix (taddhitapratyaya) to a noun, the resulting word designates a relation. For example, "aupagatva (nye ba'i ba lang gi bu nyid)" (UPAGU + AJ.l + TVA) designates the relation between an ancestor and his descendants (apatyapatyisambandha; bu rgyud dang bu rgyud dang ldan pa'i ' brei paY, in this case the relationship between Upagu and his progeny. The above observations of the effect of adding the abstraction suffix to words formed in various ways are general rules to which there are exceptions. In his commentary linendrabuddhi illustrates the three classes of exceptions outlined in the grammatical dictum. First, some nominal compounds have an idiomatic sense that is independent of the meanings of the words compounded. One such idiomatic expression (ruq.hi; grags paY is "gaurakhara (lei rtsub nyid), " which is a compound made of "gaura" meaning white and "khara" meaning rough. But "gaurakhara" as a compound is the name of a kind of wild donkey, so when the abstraction suffix is added to it, the result is a word that names not a relation but an ordinary universal, namely, donkeyhood. Second, some words formed from verb roots and primary suffIxes are also idiomatic. One example is "carin" (CAR + IN), which means human being; adding TVA to this word yields "caritva (' gro ba'i ngang can nyid)", which names the universal humanity rather than a relation between the action of going (CAR) and a factor in that action. Third, some words formed from nouns and secondary suffixes are idiomatic. An example is "hastin" (HASTA + INi [--> IN], which literally means hand-possessing but idiomatieally means elephant; when TVA is added to this word the result is "hastitva (lag ldan nyid)", which names the universal elephanthood rather than the relation between an owner and his property. These three examples give exceptions to the general rule that the abstraction suffix added to compounds, words derived from primary suffIxes, and words formed from secondary suffixes yields new words that name relations. The second exception to the general rule is a class of words for which the abstract form is formally identical with the non-abstract form. An example of such a word is "sukla," which names both a particular instance of white and the colour white as a universal property. When TVA is added to such a word, the resulting word, such as "suklatva" names a quality (gUlJa) rather than a relation. 41 The third class of exceptions to the general dictum of the effect of adding the abstraction suffIx contains words that are formally made up of the abstraction suffIx and designate an invariable relation. Some relations can be broken without altering the essence of the relata. A man may own a donkey for a while, and during this time they are in the relationship of owner and property, but if the man sells the donkey,

272

Pramaf}.asamuccaya 5:9

On linguistic signs

then this owner-property relationship comes to an end, but the man c0ntinues being a man and the donkey continues being a donkey. But other relationships cannot be broken without altering the essence of one of the relata. For example, a jug that is related to the property reality cannot be divorced from that property without also ceasing to be a jug. A property from which an object cannot be divorced without ceasing to be the same sort of object is called a universal (jtiti) and a relationship with a universal is a type of relationship in a class by itself. The grammarian Patafijali quotes the following definition of a universal: Reflective men regard a universal as that which attends the coming into being and the passing away of a particular entity and coexists with its qualities and resides ina pluraIi!y of individuals without partaking of their distinctions of gender.42 Many words with abstraction suffixes name such universals rather than ordinary relations and so form another exception to the general rule that abstraction suffixes yield words that name relations.43 Ditmaga's purpose in quoting the dictum here is to show that no matter how we construe the expression ''jtitimattva'' ("being a universal-bearer"), which is formed from the word JATI plus the secondary suffix MATup plus the abstraction suffix TVA, it is expressing, according to conventional Sanskrit grammar, either a relation or a universal.

10.0.0 How can that meaning be general? The word itself is the thing that objects have in common. But the word is not accepted without a warrant for application. Even if the universal resides in a single locus, 11.0.0 just as the property blue does, what you say cannot be the case. Even if granted, it is not relevant, because a universal does not have universals. Even in indirect implication there is ambiguity. A word excludes others. 10.1.0 "Suppose the meaning of a general~erm is an instantiation such as a jar that does not occur in other things such as cloth. How can that meaning be general?" To be general is to occur in a plurality of things. But if there is a jar that possesses a general feature and it does not occur in cloth and so forth, then how can it be general? It cannot. 10.1.1 "Suppose the meaning of a general term is an instantiation such as ajar that does not occur in other things such as cloth. How can that meaning be generalr' This passage is quoted by both PiirthasiirathimiSra and Bhattaputra J ayarnisra in their respective commentaries to Mfmiilhstislokavtirttika 131. It is noteworthy that the two Tibetan translations vary from one another in exactly the same respect that the two Sanskrit citations vary. Kanakavarman's Tibetan parallels PiirthasarathimiSra's citation.44 Vasudhararak~ita's Tibetan, on the other hand, parallels Jayamisra's

Chapter Seven

PramaIJasamuccaya 5: 10

273

reading. 45 The two Tibetan translations also differ in how they construe the syntax of the verse. Kanakavarman's rendering could be translated "If the meaning of a general tenn is an instantiation such as a jug, then it does not apply generally [to several jugs]. So how can that instantiation be the meaning of a general tenn?" In his explanation of this verse, I ayamiSra says: When it is accepted that the word "cow" refers to the subspecies Sabaleya which is qualified by cow hood, then, since that Sabaleya does not occur in the subspecies Bahuleya, the word "cow" would not apply to that Bahuleya. Despite these minor differences in the reading of Diimaga's verse, the central point is clear enough. A general tenn names a generic property, but a particular that possesses that generic property is not itself a generic property, therefore it cannot be what is named by a general tenn.

10.2.0 How can one impose this general term upon things that lack a common feature? Because, on the hypothesis that the word "real" expresses a common feature, since reality has no reality, it is not admitted to be the locus of this common feature. Therefore, one must necessarily admit that what an object has in common belongs to the particular object. But it is not in the object. Therefore, "the word itself is the thing that objects have in common." The instantiations such as the jug and so forth are similar owing to their being expressible by the word "real," but not owing to any intrinsic property of the objects named. The universal reality and the relation between it and its instantiations have already been ruled out as expressible things that have individual objects in common. 46 10.2.1 "How can one impose this general term upon things that lack a common feature? " The key issue taken up for discussion in the following passage is, as Iinendrabuddhi points out (P281b7), that DiI'maga denies that among the intrinsic properties of a particular is the fact that it has something in common with another particular. This view is expressed and argued in PramiilJasamuccaya 2:16 and in Diimaga's own commentary thereon and fOllows from his views expressed in the fIrst chapter regarding the nature of a particular and his contention that it is the only object of sensation. For in DiI'maga's system the tenn "peculiar attribute" (svalak$alJa) is restricted to sense data such as shape, colour, taste and so on that are immediately present to the senses. These sense data are the ultimate particulars of experience. Thus, for example, when a certain coloured shape is present before the eye, one straightaway becomes aware of the colour and the shape as they are, for they are intrinsic to the object sensed--or, more properly, they are the object sensed. But one need not be aware of the fact that these sensed data are in any way similar to the other objects that one has sensed or that one might sense at a later time. For, in this view; if a sentient being were to live for only five seconds and were to see in her life only one thing, say a blue circle, her cognition of what she saw would not differ substantially from what an octogenarian would be aware of in seeing his one thousandth blue circle. That the octogenarian is aware of a similarity to, or even a subtle difference from, nine hundred ninety-nine other blue circles he has seen is not derived primarily from the blue circle that is, by pure happenstance, the one

274

PramalJasamuccaya 5: 10

On linguistic signs

thousandth he has seen. Rather, it is a function of a complex m~ntal act that involves, among other things, memory. Therefore, that a datum is similar or that it may at some later time be recognized as similar to another datum is extrinsic to the datum itself. Now if there are several distinct objects, none of which has among its intrinsic properties the fact of reseinblance to something else, how can we justify applying the same general term to all of them? A general term is, after all, supposed to name a similarity or resemblance among objects. It is to this question that DiIi.naga addresses himself in 10.2.0.

10.2.2 "Therefore one must necessarily admit that what an object has in common belongs to the particular object. But it is not in the object." The view under consideration here, according to Jinendrabuddhi (Pl8Ia) is that the phenomenon of grammatical agreement or co-referentiality (siimiiniidhikaralJya) of the sort that we find in the phrase "real jug" is unintelligible unless we accept that there is a universal and its instantiation and a relation between them. Now in this view the word "real," which can be found in grammatical agreement with a variety of words such as "jug" and "cloth," names some similarity or common property. But, reasons Diimaga, if there is a similarity or common property, then it must belong to or be located in something, for this is the principle to which those who believe in universals are committed. Now the only candidates that suggest themselves for being what it is that owns this property are the universal and the particular. But a universal cannot be the locus of another universal or of itself, for the only locus of a universal is a particular substance (dravya). Thus the universal reality, which is supposed to be named by the word "real," cannot be the locus of the reality that is the feature common to the jug and the cloth and on the basis of which the word "real" is found in grammatical agreement with the words "jug" and "cloth." On the contrary, the universal reality is that which the pot and cloth supposedly have in common, and this universal is said to belong to the particular objects that are instantiations of it. This is the only position to which those who believe in real universals can take, according to Diimaga. "But," says Diimaga, "it is not in the object." As has already been mentioned above in 10.2.1, the universal, or the similarity between objects, is not regarded by Diimaga to be an intrinsic property of the objects themselves. Rather, universals, resemblance, similarity and so on are regarded as conceptual structuring (kalpana) imposed by memory upon the objects of sensation. 10.2.3 "The instantiations such as the jug and so forth are similar owing to their being expressible by the word 'real: but not owing to any intrinsic property of the objects named." Although the arguments that Dinnaga adduces in support of this position differ from the arguments mentioned by Panayot Butchvarov as being behind the nominalist theory of universals in European philosophy, one is struck by the similarity of the position under discussion here, that what one object has in common with another is just the fact that the same word is applicable to both objects, with the nominalist position as outlined by Butchvarov: According to the Resemblance Theory of universals, the instances of a recurrent quality are distinct particular qualities related by a relation of resemblance. According to the Identity Theory, they constitute an identical quality which is present in distinct individual things at the same time. According to the Nominalist Theory. they are related only Uy the fact that they are objects of the applicability of one and the same general word. 47

Chapter Seven

PramafJasamuccaya 5:10

275

10.3.0 .But one might object "the word is not accepted without a warrant for application." A word is not observed to be the same for different particulars without some warrant. And because of that, there is no relation between a qualifier and that which is qualified. But suppose we say that the word "real" does not apply to any object other than that which possesses the property reality; if after the word "real" another word [such as "jug"] is required to complete the grammatical sense--as in the case of the word "blue" which requires a suffix showing degree of comparison, forming a word such as "bluer"--then there would be a qualification relation. 10.3.1 According to Jinendrabuddhi (P282b3 ), in this and the following passages up to 11.2.0, Diimaga outlines a serious weakness of the hypothesis outlined in the preceding passage, namely, that it fails to offer any criterion for a word's being applicable to a given object or range of objects. Under the view of the Sanskrit grammarians and the Naiyayikas and others, a word is applicable to a particular object only if that object has a universal property that is named by the word in question; if an object has a plurality of properties, then a plurality of words are applicable to it. In a sentence, a string of words that have different intensions may be juxtaposed to refer all to the same object. But under a view that denies universals either as recurring or resembling properties and hence denies the intensions of words, how can one account for strings of words that are co-referential, and how can one say that one word qualifies another? Having posed this problem, Diimaga then entertains a hypothetical solution. Suppose we say that two words, WI and W2 , are in a qualification relation in case the meaning of WI would not be complete without W2 or vice versa. In this case the use of WI requires the use of W2 . Diilnaga then examines two models for this sort of requirement: a) the model of affixes indicating degree of comparison that are applied to adjectives to complete their meaning, and b) the model whereby WI alludes to or implies the object of W2 and thus requires W 2 to make this implied object explicit.

11.1.0 "Even if the universal resides in a single locus,48 just as the property blue does, what you say cannot be the case." It is impossible that the word "blue" have an instantiation that is no different from the blue object, and it has already been shown that the meaning of a word is neither blueness nor the relation of blueness to its instantiation. Moreover, "even -granted, it is not relevant, because a universal does not have universals. ,,4 Suppose it were granted that there is a common property blue belonging to that which is bluer [and bluest]. In such a state of affairs, since this common property would include internal distinctions, then, when it resided in a subject, a particular such as a jug would be required to specify it. But the universal reality does not possess other universals such as jughood and so forth in the same way that the quality blue has three degrees of

it

276

PramdfJasamuccaya 5: 11

On linguistic signs

comparison. Therefore, the example of the word "blue" and its'degrees of comparison should not be considered in this context. 11.2.0 Now one might argue that the word "real," having as its meaning the reality of things such as jugs that are not directly expressed by the word "real" but are alluded to by it, is necessarily connected with something such as jughood belonging to those objects alluded to by the word. Therefore, when one hears the word "real," one would anticipate particulars. That will not do. "Even in indirect implication there is ambiguity. ,,50 A word alludes to an object with reference to which certainty arises. For example, given that Devadatta does not eat food during the day, one is certain that he eats during the night. But here, when one says "real," there is ambiguity with reference to any given particular such as a jug, and therefore that word does not indirectly imply particulars. 11.2.1 The view under consideration here is that a general tenn implies, by a process called arlhiik(fepa, a particular and that a specific tenn simply makes that particular explicit. Thus the general tenn "real" implies some particular object that is real, and a more specific tenn such as "jug" makes that particular explicit. This account, however, is not satisfactory to Diimaga. For while it may be true that a general term implies that there is a particular to which it applies, it cannot imply any given particular. Judging from the standard examples of arlhak:fepa in the literature, with which the example given here by DiI'maga is isomorphic, it appears that the understanding of arthiik$epa is that its structure is that of what in European logic is called modus tollendo ponens or disjunctive syllogism, the logical form of which is {[(p v q) & -p] --> q}. From the unstated premiss "Either Devadatta eats during the day or Devadatta eats during the night," and the explicit premiss "Devadatta does not eat during the day," we may conclude "Devadatta eats during the night." Now if a general term W were to yield knowledge of a particular PI' or in other words, one particular from the set of particulars to which W is applicable, and if this knowledge were derived from a fonn of reasoning that had the same logical form as the disjunctive syllogism, then the logical structure of that general term as an inferential sign would be somewhat like the following: [(W names PI) v (W names Pz) v (W names P 3) v ..• (W names Pn)] & [-(W names P2) & -(W names P3) & ... -(W names P n)] --> (W names PI)'

But if the general term is applicable to the set of particulars {PI' P 2 , P3 , ••• Pn }, then it is not the case that it does not name one of {P2"'Pn}' Hence the general term alone cannot pick out just one particular from the set of particulars to which it is applicable. A .general tenn, therefore, does not function by the process of arthiilqepa or disjunctive syllogism.

11.3.0 Since there is no way that a general term can express particulars (bheda; khyad par) or universals (jdti; spyi) or the relations between particulars and universals or instantiations of a universal, "a word excludes

Chapter Seven

PramalJQsamuccaya 5: 11

277

others. n Therefore, we have established the truth of what was said at the

outset of this discussion, namely, that a word expresses its own object by precluding what is incompatible in a way similar to [an inferential sign such as] the property of having been produced [indicates a property such as the fact of being transitory].

12.0.0 Although that which is expressed by a word has many properties, it is not cognized in its entirety through a word. The word performs the task of isolating its referents according to its intrinsic relations to what it expresses. 51 13.0.0 A word too has many properties. But it makes its object known only through those properties by which it does not deviate from the object; it does not make its object known through the fact that significant sound is a quality and other such properties. 52 13.0.1 Frauwallner pointed out the similarity of this verse to PramiifJasamuccaya 2:17.53 In his commentaries to PramiiQasamuccaya 2: 13 and 2: 17, Diimaga explains that an inferential sign such as smoke can convey knowledge of only those properties of the sign bearer, such as fife, that occur at every locus of fire. Smoke can make known only those properties without which no fire exists. But such properties as, for example, being red or having a given specific temperature, which are properties that occur at only some loci of fife, cannot be known through an inferential sign. As for properties of the inferential sign, only those that are restricted to the loci of the sign can playa role in making the sign bearer known, but properties whose extension is wider than that of the property serving as an inferential sign cannot playa role in making the sign bearer known. Thus smoke's having a smoky colour, which property is restricted to smoke, can play the aforementioned role, but smoke's property of being a substance cannot, for being a substance is not restricted to smoke. Now in this present chapter, Diimaga asserts that a linguistic sign or word functions in the same way as a property used as evidence. Thus all that has been said of smoke in the discussion above can now be said of the word "fife." Only those properties of the word "fire" that are restricted to that word can make fife known, and of course the only things about fire that the word "fire" can make known are those properties without which no fife exists. Thus for example, according to some schools of Indian metaphysics, sound is a quality of ether. But this fact of sound's being a quality plays no role in making fire known through the articulated sound "fife." What does playa role in making fife known is the property that the word "fife" has of not being applicable to any state of affairs in which fire is absent And in this rtspect the word "fire" is said to be exactly parallel to smoke, which makes fife known by virtue of its never occurring in any locus in which fife is absent.

278

Prama;;asamuccaya 5:14

On linguistic signs

7.2 On the relationships between symbols that express precliIsion

14.0.0 Words that have different meanings owing to a difference in what they preclude are inadequate with respect to conveying knowledge of the particulars that fall within their respective extensions. There is a qualification relation between two words owing to their having the same effect when both are applied to the same thing. 14.1.0 If the meaning of an expression is nothing but preclusion of the complement of the set of things to which the word applies, then it would seem that there can be neither co-referentiality nor a qualification relation in such expressions as "blue lotus." Why not? Because the general term and the particularizing expression preclude different things. But in fact this problem does not arise in our theory. "Words that have different meanings owing to a difference in what they preclude are inadequate with respect to conveying knowledge of the particulars that fall within their respective extensions. There is a qualification relation between two words owing to their having the same effect when both are applied to the same thing."s4 Although the words "blue lotus" do differ in the complements that they preclude, the words are in grammatical agreement in virtue of the conjunction of the respective complement-preclusions in one object. This conjunction serves the purpose of pointing out particulars in each word's extension, like a crow standing on a house. Thus each of the two words by itself is a source of uncertainty concerning the objects it refers to. Moreover, since a single word's object cannot be the object that is made known when that word is accompanied by another word, there is a qualification relation. 14.1.1 The seven verses PramiifJasamuccaya 5:14-20 deal with the related questions of grammatical agreement as based upon co-referentiality, whereby two or more terms with different intensions can refer to a common extension, and the relation of qualification, whereby an attribute is predicated of a subject. Diimaga's strategy is to establish first of all that these two phenomena can be accounted for in his theory, then to show that they cannot be adequately accounted for in other theories of meaning. Thus the first question to which he addresses himself is whether under the apoha theory of meaning two words that have different intensions can be said to apply to the same set of objects, given that the intension of a term in this theory is the preclusion of its counter-extension.

14.1.2 "This conjunction serves the purpose of pointing out particulars jn each word's extension, /ike a crow standing on a house." The simile of the crow standing on a house occurs in Helaraja's commentary to Bhartrhari's Viikyapadlya 3.2.3, where,

Chapter Seven

PramiilJ.asamuccaya 5:14

279

although it is used to make an entirely different point, the general meaning of the simile is explained clearly enough to allow us to see how it is being used here. The general point is that a word or other sign can be used to point out something even to which it is related only accidentally, just as one might indicate a particular house to someone seeking its whereabouts by saying "It is that house over yonder with the crow sitting out front." The particular application of that simile to this passage is that one word WI can isolate objects in the extension of another word W 2• despite the fact that WI has no intrinsic relation with the extension of W 2 . Any property divides the universe into two disjoint classes, namely, those things that have the property in question and those that lack it. A word that names a property excludes those objects that lack that property, and so it divides the universe into the class of things to which the word is applicable and the class of things to which the word is not applicable. If the two words are juxtaposed to form a larger unit of expression, then each of the words will exclude some of the items of the other's extension. Thus the word "blue" functioning as a part of the expression "blue lotus" eliminates from the extension of the word "lotus" all those things to which the word "blue" is not applicable, while the word "lotus" eliminates from the extension of the word "blue" all those blue things to which the word "lotus" is inapplicable. In principle any two words can be in a relation of grammatical agreement or samaniidhikarar;ya so long as neither eliminates all of the other's extension, that is, so long as the intersection of their extensions is a non-empty set. Furthermore, since the extension of the word WI is reduced in size when WI is juxtaposed with another word W2 , we can say that W 2 qualifies WI' In the same sense, WI can be said to qualify WI' For WI alone will have a larger extension than that of WI W2 combined, and W2 will also have a larger extension than that of WI W2 combined. Therefore it is the contention of Diimaga that in a qualification relation between words, each word qualifies the other.

15.0.0 Since the thing expressed is a complex entity, neither blue in isolation nor a lotus in isolation is expressed. [The individual words] have no meaning, just as the individual phonetic components of a word have no meaning. 15.1.0 But given that that which is expressed by the compound "blue lotus" is neither the quality blue nor the substance lotus, how is it that the expression "blue lotus" expresses a single object, namely, one that is both blue and a lotus? "since the thing expressed is a complex entity, neither blue in isolation nor a lotus in isolation is expressed." From the two words, "blue" together with "lotus," one is aware of the two things combined, but one is not aware of them individually. The two words functioning as elements of a complex expression individually "have no meaning, just as the individual phonetic components of a word have no meaning."55 The situation here is like that of the word "nlla" (="blue"),

280

Prama/J.asamuccaya 5:15

On linguistic signs

which is meaningful despite the fact that the individual phonetic components "nt' and "la" are meaningless. 15.1.1 "How is it that the expression "blue lotus" expresses a single object?" This question possibly anticipates an objection to Diimaga's apoha theory that might be raised by someone who subscribes to the theory that individual words names individuals and that strings of words taken together as compound expressions name something like the class of things in which all the universals named by the individual words inhere. Under this theory the word "blue" in the expression "blue lotus" has as its warrant of application (pravrttinimitta) the universal blueness or the quality of blue colour, the word "lotus" has as its warrant of application !he universal lotushood, and the warrant for the juxtaposition of the two words is the collocation of these universals in a single locus or set of individual loci, which is expressible by the compound expression. The challenge is now put to Diimaga to account for the meanings of compound expressions under his theory that denies that words name positive entities such as universals.

15.l.2 Ruegg has given a concise account of the history of the question raised by classical Sanskrit grammarians as to whether or not individual syllables or letters in words convey any meaning: Une question fondamentale se pose a ce point: les lettres (van;a) qui constituent Ie mot sont-elles signifiante (vartt 9 ad Sivasutra 5: arthavanto varfJii dhiitupriitipadikapratyayanipiitiiniim ekavarfJiiniim arthadarsaniit, etc.)? Puisqu'une signification ne resulte pas de l'union d'un rassemblement que1conque de lettres (varfJasanghiita) et puisqu 'un suffix ne peut etre affixe a une lettre seule, on conclut que les lettres individuelles ne sont pas signifiantes (vartt. 14: aoorthakii tu prativanjam arthiinupalabdhel;!), a mains, bien entendu, que la lettre ne forme a elle seule une unite signifiante, un mot. II est done dit que les lettres sont a la fois sans significations et signifiantes ubhayam idmn varfJe:;aktam.f6 Ruegg goes on to explain how Bhartrhari concludes that distinctions of words within sentences and letters within words are artificial and imaginary.57 The sentence alone is, according to Bhartrhari, significant. On this view the individual words function in a sentence not to name an object directly but rather to reveal the sentence itself, and it is then the full sentence that names an object, namely, a conception of a whole state of affairs. Ditmaga at PramiifJGsamuccaya 5:46-49 and in his own commentary to 2:4d endorses this theory of Bhartrhari's. Now given this view that sentences rather than their component parts have meaning, it seems that Ditmaga could treat a compound expression like "blue lotus" in one of two ways. First, he might regard it as a kind of new word, one that is distinct from its component parts because the extension of the compound differs from the extensions of its component parts. The most radical form of this view would hold that the fact that the new word "new lotus" appears, but only appears, to be made up of parts that are themselves meaningful could be regarded as a sort of phonetic accident similar to the fortuitous fact that the word "migrate" sounds as if it is made up of the two separate :.vords "my" and "great." Such ambiguities of sound are harmless enough

Chapter Seven

PramalJasamuccaya 5:15

281

when they occur as parts of sentences, for one who knows English reasonably well is unlikely to be in doubt as to which of these two phonetic configurations ought to fit into the slot of these sentences: "Geese .... to Canada in the summer" and "This is an heirloom from ... grandfather." But the fact remains that most languages abound in examples of such fortuitous coincidences of phonetical shape in units of speech shorter than the sentence. If we take Dmnaga's analogy quite literally, this interpretation seems plausible, for he does say that "nlla" and "utpala" in "nflotpala" are no more meaningful than. nI and "la" in "nIla". And under this interpretation it would follow from Dinnaga's theory of the functions of words in sentences that the word "blue-lotus" would itself, in the context of a sentence such as. "a blue lotus is in the pond," cease to be a meaningful unit of expression and would be relegated to the function of revealing the object-naming sentence of which it is a part. But a second interpretation of the compound "blue lotus" is possible within the framework of Dinnaga's theory, namely, that it is itself an abbreviated sentence, something akin to an atomic proposition. An atomic proposition in this context is a simple predication which is capable of being embedded in a larger sentence that is, despite'its appearance on the surface as a grammatically simple sentence, in fact a logically complex sentence. Under this interpretation a sentence like "A blue"lotus grows in the mud" could be regarded as a concise colloquial form of the logically compound sentence "There is a lotus that is blue, and it grows in the mud." The compound "blue-lotus" then is a shortened form of the atomic sentence "lotus is blue."S There are, unfortunately, insufficient data on which to rest the case for either of the two possible interpretations outlined above, for it is not clear whether Dinnaga's criteria of what counts as a sentence are the purely formal criteria of the grammarian or the logical criteria of the logician. All that is clear for the example at hand is that Dinnaga has said that the individual words in a nominal compound are as devoid of meaning as the individual syllables in a word. In 16.1.0 he anticipates an objection to this view.

16.0.0 Although the individual phonetic units have no meaning, each of the two words in a complex expression do make their respective objects known. Supposing you say the word does express something, even in this view what it makes known is a thing in isolation. 16.1.0 One might argue that this [comparison of the words in a compound to the syllables in a word] is a false analogy. When part of a word is pronounced, one does not understand any meaning, but when the word "blue" is pronounced all together, one does understand. "Although the individual phonetic units h4ve no meaning, each of the two words in a complex expression do make their respective objects known." [I reply] "Supposing you say the word does express something, even in this view what it makes known is a thing in isolation." The words "lotus" and "blue" taken separately are devoid of the meaning of the compound expression "blue lotus" in the same way that the phonetic components of the word "nfla, " namely, the syllables "nf'/ and "la," are devoid of reference to any external object. Even in the view that says some meaning is understood

282

PramalJasamuccaya 5: 16

On linguistic signs

through the word "blue," since what is expressed by the word "blue" is a universal alone, a thing in isolation is cognized. But that universal blueness that is expressed is connected with a particular quality and with a substance. Therefore, the word "blue" is as devoid of the meaning of the entire complex as its phonetic components are devoid of the meaning blue. Hence it is the compound expression consisting of words as devoid of meaning as their phonetic components that make the entire complex of substance and all that inheres in it known. We say that the meanings of different expressions differ by virtue of the objects expressed. Therefore, it is possible that words of which the meaning is preclusion of objects other than those to which the word is applicable can be in gramillatical agreement and in a relation of reciprocal qualification. 16.1.1 According to those who, like the Vaise~ikas and others, agree that a word such as "blue" expresses a universal, the universal that is expressed can be cognized only through the medium of particular instances of it. In the case of "blue," it names the universal blueness alone. But this blueness inheres in a particular instance of blue, which is a quality that in turn inheres in a substance. But neither the substance in which the particular instance of blue inheres nor the particular instance of blue in which blueness inheres is expressible by the word "blue," because the word "blue" is not restricted to any substance nor to any particular instance of the quality blue. Therefore, argues Difmaga, it must be conceded that the word "blue" is as incapable of expressing the entire reality of what the compound "blue lotus" expresses as a phonetic component of the word "blue" is incapable of expressing what the word "blue" expresses. In order to express the full reality of blue lotus, the entire expression "blue lotus" is required. Diimaga' s line of reasoning here does not, Ot course, really address the issue squarely, for the opponent would not argue that "blue" in "blue lotus" expresses all the meaning of "blue lotus" but rather that it retains in a compound at least some of the meaning it had as an independent word and that it expresses part of the meaning of the compound expression.

7.3 On the unreality of universals outside thought

17.0.0 A unified complex entity does not exist, since it would then follow that the components are identical with one another [and] because it would then follow that the complex entity would be many. They do not give up their individual meanings. 17.1.0 A word's meaning cannot be anything other than preclusion. Why? A substratum must be either identical with or different from its component parts. Considering the first of these alternatives, it cannot be identical, because "A unified complex entity does not exist, since it would then/ollow

Chapter Seven

Pramiil;asamuccaya 5: 17

283

that the components are identical with one another." If the complex entity were a unity, then the two objects blue and lotus would not be different from it. Therefore, since they do not differ from the unity, they would not be different from one another. Moreover, a unified complex entity does not exist, "because it would then follow that the complex entity would be many." Since the complex entity is not different from the several components making it up, its plurality would follow from the thesis of identity. Therefore, it does not exist. But even if the existence of a complex entity is accepted, the two objects blue and lotus cannot have a single substratum, because even when both words are applied to a single object, "they do not give up their individual meanings." The meaning of each of the individual words "blue" and "lotus" is its own universal, and this remains true when they are in a compound expression. Therefore, how can they be in grammatical agreement? 17.1.1 In accordance with the strategy outlined above in 14.1.1, Diimaga, having argued that grammatical agreement and the qualification relation can be accounted for under the theory that words express preclusion (apoha), goes on to argue that they cannot be accounted for under the theory that words express universals. Under this theory two or more words may be said to be in a relation of grammatical agreement or co-referentiality (samanadhikarar;ya) only if those two words name universals that inhere in the same substratum. Thus grammatical agreement is regarded as a linguistic reflection of an ontological fact, namely, the inherence-in-a-singleparticular (samantidhikarar;ya) by more than one universal. In his critique of this theory, Difmaga begins with an examination of the nature of that particular that serves as a substratum for a plurality of universals. In the discussion that follows he attempts to show that undesired consequences result whether the particular substratum be regarded as identical with its component parts, or whether it be regarded as different. In the context of this discussion, the component parts of a thing are the several universals that inhere in it. The "complex entity" (bums pa; *samudaya) in question here is that particular object, such as the blue lotus, that is hypothetically regarded as the substratum of a plurality of universals, such as blueness and lotushood, and as the denotation of a compound verbal expression such as "blue lotus." For Diimaga such a complex entity is the referent of a compound expression, but unlike his opponents he does not accept it as an extramental reality or as an object available to sensation (praryakya).

18.0.0 It is not the universal, because it was explained above. Because of uncertainty. Because the word does not exclude that object. That which is made known through synonyms is made known through only one of the set of synonymous expressions. If the whole is different from its parts, both grammatical agreement and the qualification relation are contravened. 18.1.0 Now it could be thought that that [observation in 17.1.0] poses no problem. One could say that both words have as their meaning their

284

Pramar;asamuccaya 5: 18

On linguistic signs

respective particulars; nevertheless, since particulars are included in classes, the two words have a feature in common owing to there being a set of objects common to both classes. Both words are used in order to make those particular objects known. Thus the two words may be in grammatical agreement because they have a complex entity as their meanings. That will not do. First of all, that which the word "blue" expresses as non-different [from that which "blue" names] "is not the universal," because the substance-universallotushood is considered to be inherent in the particular blue lotus, but the word "blue" does not invariably express that substanceuniversal. Now one might suggest that it does not invariably express a substance endowed with the quality blue, and the particulars of that class of substances endowed with the quality blue, such as blankets and flowers, would be included [within the extension of the word "blue"]. That is untenable. Why? "Because it was explained above." For the same reason as was given above [beginning at 4.2.0], "[a general term does} not [express} an instantiation; because it [scilicet the word that expresses an instantiation} is grammatically subordinate," we argue here against the word "blue" expressing an instantiation of the quality blue. Similarly, he is incorrect who argues that there is a relation of grammatical agreement because particulars are included in genera. Why? "Because of uncertainty." We observe that there is uncertainty from a general term with respect to the particulars that it is being applied to. That from which uncertainty about a thing arises does not express that thing. A wider class may be cognized immediately through a narrower term because it [scil. the extension of the narrower term] is never errant [from the class in which it is contained, namely, the extension of the wider term]. 18.1.1 When two words are in apposition, the relation of the words is conventionally. regarded as a reflection of the fact that the objects narned by the two words are identical to one another. Now in the expression "blue lotus," the question arises as to whether each of the two words expresses a universal or the instantiations of a universa1. 59 In other words, does "blue lotus" assert identity of blueness and lotushood, or does it assert that instances of blue are identical with instances of lotushood? Diimaga rules out the fIrst of these alteruatives with the phrase "First of ail, that which the word 'blue' expresses as non-different [from that which 'blue' names} 'is not the universal'. " The second alternative is ruled out with '''Because it was explained above.' For the same reason as was given above '[a general term does] not [express} an instantiation; because it is grammatically subordinate'." By the same argument, of course, "lotus" cannot be regarded as expressing instances of lotushood. And so the juxtaposition of "blue" and "lotus" cannot be construed as expressing the identity of instances of blueness with instances of lotushood.

18.1.2 "Similarly, he is incorrect who argues that there is a relation of grammatical agreement because particulars are included in genera." What is not clear in this passage is what sort of misconception Diimaga is here taking pains to rectify. Iinendrabuddhi offers no comment on this passage, so its meaning was either so

Chapter Seven

Pramdr;asamuccaya 5:18

285

obvious as to require no comment or so obscure as to make offering a comment too risky. What seems to be at issue here is some view according to which if smaller classes are contained in larger classes, then knowledge of smaller classes is part of our knowledge of larger classes. On the passage "A wider class may be cognized immediately through a narrower term because it is never errant," linendrabuddhi gives the illustration of the word "sbh$apa," from which one can know that the object to which it is applied is not outside the class of trees. One knows this because of the fact that anY,object to which "sirhsapa" is applicable is an object to which "tree" is applicable.oO From the word "tree," on the other hand, one cannot know whether the object to which it is applied is within or outside the class of things to which "sirhsapa" is applied. Thus one can say that "sirhsapa" expresses the fact of being a tree but not that "tree" expresses the fact of being a sirhsapa. The principle that one may reason from a thing's membership in a narrower class to its membership in a wider class but not vice versa forms the heart of PramalJasamuccaya 2:12-25. The same principle underlies the present passage, which parallels the discussions in 2.1.0.

18.2.0 It might be pointed out that there is an authoritative text that says that particulars are included in genera. That notwithstanding, it is not the case that a word expresses an object owing to that object's being expressible by the word, but rather "because the word does not exclude that object." Since the word "blue" does not exclude the sesame seed as it does the jasmine flower, the sesame is included in that class of blue things. Even if one accepts this account, the sesame is included among the objects expressible by the word "blue." Moreover, if both words "blue" and "lotus" have the same particular as their object, then the two words would be synonymous. And we observe that "that which is made known through synonyms is made known through only one of the set of synonymous expressions." One does not cognize more than one object through a pair of synonymous expressions, but a second synonym is applied for the sake of making known the object of the first one. This is how synonymous words are by nature. Synonymous terms do express their objects, but not at the same time. But in that case, since the object lotus would be cognized through the word "blue" alone, there would be no need for the word "lotus" and therefore there would be no relation of qualifier and qualificand. Thus there is a problem in the view that the substratum of the two words "blue" and "lotus" is identical. 18.2.1 Diiulllga's purpose here is to emphasize his view that the word "blue" is applicable to the sesame and the lotus not because the sesame and the lotus have in common a property that "blue" names, but rather because nothing prevents our applying the word "blue" to the sesame and to the lotus. Thus, it is not so much a matter of sesame and lotus being members of the class of blue things as of there being no contradiction involved in our saying "This is both blue and a sesame." Concerning the passage "One does not cognize more than one object through a pair of synonymous expressions, but a second synonym is applied for the sake of making known the object of the first one," linendrabuddhi (P294b1) offers several examples

286

Pramiil;asamuccaya 5: 18

On linguistic signs

of words that have a variety of applications. In such cases it is commo!1 practice to provide a gloss to specify which of the several applications of the ambiguous word is intended. But this gloss does not name a further object in addition to that named by the word to which it is a gloss. But, so the argument goes, if "blue" and "lotus" are construed as simply naming the same particular, then, since "blue" and "lotus" are co-extensional, they would be synonyms. In that case, "lotus" might be regarded as a gloss for "blue" in about the same way that in the expression "He is a gentleman, a man of noble birth," the phrase "a man of noble birth" names nothing in addition to what "gentleman" names but serves only to clarify in which of its various allowable sense "gentleman" is being used. But if "blue" and "lotus" are synonymous, then neither word is qualifying or modifying the scope of the other.

18.3.0 But there is also a problem in the view that the substrata are different. It must be shown that the whole is something other than its parts. But this cannot be shown, because neither wholly nor partially can the whole reside in the parts or the parts in the whole. But even if it is accepted, that is, "if the whole is different from its parts, both grammatical agreement and the qualification relation are contravened." In this case too, grammatical agreement and the relation of qualifier and qualificand turn out to be impossible. The two putatively distinct relata may be regarded either as two objects or two expressions. If they are two objects, then either they are a quality and a universal or a quality-bearer and the instantiation of the universal. Similarly, if the two relata are two expressions, they are either that which expresses a quality and that which expresses a universal or that which expresses a quality-bearer and that which expresses the instantiation of a universal. These are the alternatives to be examined. 18.3.1 The relations known as stimtintidhikarafJya and visefjar}avise:;yabhtiva are traditionally considered to obtain both between words and between objects. At the level of language, if two words are. in the stimtintidhikaralJya relation, they are coreferential or in grammatical agreement. What this means is that each word applies to the same object as the other. This linguistic relation is said to reflect the objective relation obtaining between two properties, namely, the occurrence of the two properties in the same locus. Similarly, a qualifier (visefjalJa) at the level of language is a term that names a specific subset of the extension of the term qualified, the qualificand (vise:;ya). This linguistic relation is said to reflect the objective relation obtaining between either genera and species or substances and qualities. In this passage, DiilIliiga examines both stimtintidhikaralJya and visefjalJaviseuabhtiva at the levels of both language and objects.

19.0.0 If two objects, a quality and a universal, subsist in a single locus, then there is no relation of genus and species. But both relations are lacki..l1g in the bearers. 19.1.0 Among those alternatives, "if two objects, a quality and a universal, subsist in a single locus, then there is no relation of genus and

Chapter Seven

Pramiil).asamuccaya 5:19

287

species." Although there is residence in a common locus owing to both the quality blue and the ,universal lotus 's residing in a single substance, there is no relation of qualifier and qualificand between the quality blue and the universal. lotushood. The quality blue does not bear the universal lotushood, nor does the universal lotus hood bear the quality blue. "But both relations are lacking in the bearers [of the quality and of the universal]." If the objects under consideration are that which bear the quality blue and the lotus-universal, then there is no residence in a common locus on their part, because neither of the two substances resides in anything other than itself. Also there is no relation of qualifier and qualificand, because the genus-species relation is untenable whether [the quality-bearing substance and the universal-bearer occur as] two separate things or as a single entity.

20.0.0 The same is true for two words expressing just the universal and the quality. In the case of two words expressing the bearers, it follows that they are both the same and not the same. 20.1.0 "The same is true for two words expressing just the universal and the quality" as was true of the two objects themselves. The word expressing a universal and the word expressing a quality differ in meaning, therefore they do not have a co-referential relation. And because the two words are not related, they also have no relation of qualifier and qualificand. Thus both relations are lacking in words expressing just the universal and the quality. "In the case of two words expressing the bearers, it follows that they are both the same and not the same." Since the two words express a single object that is qualified by a quality and a universal, they would be the same with respect to meaning. But since they express that object by expressing different qualifiers; they would not be the same. Hence it is uncertain whether or not the two words are co-referential.

7.4 On the question of what linguistic symbols preclude

25.0.0 A term does not exclude terms of wider extension, terms of narrower extension or co-extensive terms, even though they differ from it. 61 25.0.1 The content of the information that the use of a term imparts to its hearer is, according to DiI'maga, that different terms are not applicable to the object to which

288

Pramar;.asamuccaya 5:25

On linguistic signs

the tenn in question is not being applied. In this context two tenns aTe said to be different from one another if they are not tokens of the same symbol class, that is, if they differ in phonetic fonn or represent different lexemes. It is obviously not the case that a given tenn' s applicability to the object precludes the applicability of all tenns that have a different phonetic fonn. In this and the following passages Diilnaga sets out a series of observations concerning which tenns are precluded and which tenns are not precluded by a given tenn.

25.1.0 A term does not exclude the meanings of wider terms or narrower terms, even though such terms are different in form from the terms in question. Why is this? Because they are not incompatible. First, a synonymous term, which is the same with respect to what it preclud'es because it is not applied at the same time, cannot contradict that which has the same object of applicability. Similarly, the preclusion of other objects that a wider term makes for its own members is not overruled by a narrower term, because this preclusion made by the wider term is suitable to the narrower term. For example, while a sirhsapa tree is not a palasa tree, it is also not a jug. This means that a term also does not contradict a term that is wider than any term that is wider than the term in question. Thus when a wider term places its own objects in a domain that is acceptable to the narrower term, how can a narrower term or a term that is even narrower than that narrower term not condone it? Thus, since there is no contradiction involved, a term cannot exclude the meanings of wider terms. In the same way, without contradiction, a string of words can also express another object. As has been explained above, it is possible for two or more terms that are wider and narrower with respect to one another and that apply to a property common to their respective objects to express in the same way another object that is qualified by that common property. 25.1.1 "A term does not exclude the meanings a/wider terms or narrower terms." Diilnaga ultimately regards all tenns as general terms. Some tenns, however, are more general than others. Given any two general tenns G1 and G2 , we can say that one is wider than the other only if the extension of one is a proper subset of the extelision of the other, that is, only if a) G 1 is applicable to every object to which G2 is applicable but G2 is not applicable to every object to which G1 is applicable, or b) G2 is applicable to every object to whicli G1 is applicable but G1 is not applicable to every object to which G2 is applicable. Thus "oak" can be said to be narrower than "tree," and "tree" in turn is narrower than "vegetation." "Tree," then, is a narrower tenn (visel)asabda) with respect to "vegetation," but it is a wider tenn (stimtinyasabda) with respect to "oak." In the case of two general tenns whose extensions intersect without one being the proper subset of the other, such as "mathematician" and "philosopher," we cannot speak of one term being wider than the other. Similarly, in the case of two general tenns whose extensions do not intersect at all, such as "male" and "female," we cannot speak of one being wider than the other.

Chapter Seven

Pramiil)asamuccaya 5:25

289

25.1.2 "A synonymous term, is the same with respect to what it precludes because it is not applied at the same time." The test for the synonymy of two terms that DiI'maga gives here, while far from rigorous, appeals to the intuition that when two terms are synonymous, then we employ one or the other but not both to an object within a given unit of discourse, except in those peculiar circumstances pointed out. above in 18.2.1 in which synonymous terms are juxtaposed in order to disambiguate one or both of the terms. There are also cases when synonyms are juxtaposed for emphasis as in the English cliche "in any shape or form." But aside from the special cases of disambiguation and rhetorical emphasis, only one of a set of synonyms is applied in a sentence. Thus, supposing that "scholar" and "pundit" are synonymous, a speaker would say either "Sumati is a scholar" or "Sumati is a pundit" but not "Sumati is Doth a scholar and a pundit." The hearer of a sentence such as "Sumati is a scholar," knowing the synonymy of the terms, would know that the term "pundit" is not inapplicable to Sumati, even though "scholar" and "pundit" belong to different symbol classes. 25.1.3 "The preclusion of other objects that a wider term makes for its own members is not overruled by a narrower term. A term also does not contradict a term that is wider than any term that is wider than the term in question." A general term such as "tree" when applied to an object precludes the applicability to that object of any term that is incompatible with "tree." A term is incompatible with "tree" if it does not have any trees within its extension, that is, if it cannot correctly be applied to a tree. This preclusion of any term incompatible with "tree" is carried over to any term narrower than "tree," that is, to any term the extension of which is a subset of the extension of "tree." Thus "oak," being a term narrower than "tree," precludes all that "tree" precludes and more. The further things that it precludes, of course, are all the members of the subsets of the extension of "tree" to which "oak" is inapplicable, such as the extension of words such as "elm," "juniper," "sycamore," and so on. The relation of being wider is a transitive relation; that is, if a term G1 is wider than G2 and G 2 is wider than G3 , then G 1 is wider than G3 . Similarly, the non-contradiction of two terms, which is a function of one term being wider than (or, more accurately, at least as wide as) the other, is shown here to be transitive. That is, if G3 does not contradict G~ and G2c does not contradict G 1, then G3 does not contradict G 1 . What is also implIed in 25.1.0 and will be stated more fully later on is that noncontradiction is a symmetrical relation. That is, if G1 does not contradict G2 , then G 2 does not contradict G 1: The full import of the passage under consideration, then, is that given any two terms Gm and Gn , if they are coextensive or if one is wider than the other, then, regardless how many terms whose extensions intermediate between Gm and Gn there may be, Gm and Gn are not contradictory terms. The notion of hierarchy of extension is expressed by Dinnaga as follows. If term G1 is wider than G 2 and G2 is wider than G3 , then G2 is called a "samanyasabda (spyi'i sgra)" with respect to G3 and a "bhedasabda (khyad par gy.i sgra)" with respect to G 1. G 1 is called a "samanyasamanyasabda" with respect to G 3, and G 3 is a "bhedabhedaiabda" with respect to G 1. In applying the principle stated in 25.1.0 "samanyasamanyaiabda" with respect to a term G i may be regarded simply as a "samanyasabda" with respect to G r 25.1.4 '''Thus when a wider term places its own objects in a domain that is acceptable to the narrower term, how can a narrower term or a term that is even narrower than that narrower term not condone it?" In this and the following passages Dinnaga presents an hypothesis wherein words are spoken of as if they had human emotions

290

Pramar;asamuccaya 5:25

On linguistic signs

and attitudes. Thus words that contradict one another are said to show jealousy and hostility towards one another, while words that are not contradictories are said to be friendly. In the present passage, it is suggested that wider tenus place the objects to which they are applicable within reach, so to speak, of their narrower tenus. The narrower tenus, then, cannot begrudge this friendly gesture of the'wider tenus and so allow them to apply to the same objects as the narrower tenus themselves apply to. 25.1.5 As Iinendrabuddhi points out (p302a8.303a i ), it has been argued above in PramafJ.asamuccaya 5:14-16 that a string of words (tshags; *samllha) has an extension that is different from the extensions of the individual words in the string. In particular, the extension of a string WI W 2 is a subset of the extension of Wi and W 2 . Thus the expression "blue lotus," which is applicable to members of the intersection of the set of things to which "blue" is applicable and the set of things to which "lotus" is applicable, expresses a different set of objects (don gzhan; *artMntara) from the sets expressed by "blue" and "lotus" respectively. As long as the intersection of the extensions of two tenus in a string is not empty, neither of the two tenus contradicts the other, nor does the string itself contradict either of the components of the string. This principle applies not only to strings that state predications, such as "blue lotus," but also to strings that constitute grammatically complete sentences such as "The handsome royal minister approaches. ,,62 In this latter case the sentence expresses an object (vakyartha) that is the intersection of three sets, namely, the set of royal ministers, the set of things to which "handsome" is applicable, and the set of things that are approaching the speaker at the time that he utters the sentence. Since the sentence as a whole may be regarded as a general tenn that is narrower in extension than any of the general tenus making it up, the sentence does not contradict any of the tenns making it up. And since each general tenu is applicable to members of a subset of the extension of each other general term, none of the words in the sentence contradicts any other word in the sentence.

26.0.0 A wider tenn does not exclude its narrower tenns, because it creates anticipation for them alone. Neither does it entail them, because uncertainty arises concerning which of its narrower tenns is applicable. In either case the two tenns can apply to the same set of objects. 63 26.0.1 A relatively wide tenu such as "animal" creates expectation in the hearer for more specific infonuation to be conveyed by a narrower tenu, but it also confmes that anticipation to a selected domain of objects, namely, just those objects to which the wider tenu is applicable. Thus the wider tenu selects a set of narrower tenus, namely, just those tenus that are applicable to some subset of the wider tenn's extensions such as ("horse", "opossum", "cat", "rhinoceros" ... ). The wider tenu cannot, however, select any specific member of that set of narrower tenus, because it is not restricted to any of them.

Chapter Seven

PramafJ.asamuccaya 5:27

291

27.0.0 A narrower term, being restricted to more than one wider term, can entail several wider terms. There is not a symmetrical qualIfication relation between the two terms. 27.0.1 As Iinendrabuddhi explains (p303b4 -8), there are strings of symbols in which each symbol qualifies the other. A symbol S qualifies a symbol T if S is applicable only to a proper subset of T's extension. Thus in a symbol string such as "blue lotus," the word "blue" applies to a subset of the extension of "lotus" and "lotus" applies to a subset of the extension of "blue," so each word qualifies the other. In contrast to this symmetrical qualification relation, there are strings of symbols made up of wider and narrower terms in which one symbol qualifies the other but not vice versa. In a string such as "oak tree," for example, "oak" is applicable to an object only if "tree" is applicable to that object, and so "oak" applies to a proper subset of the extension of "tree." But it is not the case that "tree" is applicable only if "oak" is applicable, and so it is not the case that "tree" applies only to a proper subset of the extension of "oak." Hence "oak" qualifies "tree" but "tree" does not qualify "oak," and so this qualification relation is asymmetrical (khyad par dang khyad par gzhi

dag rnthsungs rna yin; na tuZya vise,raTJavifef)'ata).

64

28.0.0 A narrower term precludes the objects denoted by other narrower terms because of hostility. Because particulars of another wider term are contradicted by the word sirhsapa's own wider term. 28.1.0 But why does one narrower term preclude the objects denoted by other narrower'terms? "A narrower term precludes the objects denoted by other narrower terms because of hostility. ,,65 This is so because words denoting different subsets of the same set are mutually hostile, and therefore, like the sons of a king, each expropriates their common property for himself. Therefore, they begrudge one another the property they have in common. For example, in the expression "This is a sirhsapa tree," the word "sirhSapa" that is used with "tree" expropriates the treehood from the palMa and other kinds of tree and places it in its own domain. The others do likewise. Thus, flrst of all, because the narrower term takes sale possession, it can preclude objects denoted by other narrower terms. Why does the word "sirhsapii" preclude other objects such as the jug that are unrelated because they are in the extensions of different wider terms than "tree"? "Because particulars of another wider term are contradicted by the word 'simsapa" s own wider term." Since the word "tree" expropriates from the jug the property of being earthen, it contradicts the term "jug. ,,66 When the wider term "tree" precludes the jug, which is like the enemy of a friend, the narrower term "sirhSapa" is content in knowing that the jug has been precluded from the extension of "tree" and therefore of "sirhsapa." We are to understand this to mean that the term "sirhSapa," like an enemy

292

PramalJasamuccaya 5:28

On linguistic signs

of a friend's friend, precludes general terms for qualities that are particulars of other wider terms, but, like a friend of a friend's enemy, it tolerates terms for colours that are particulars of those particulars of other wider terms. Whether a term is precluded or tolerated depends on its relation to the terms related to the term that precludes it or tolerates it. 28.1.1 In 25.1.0 we find the following principle stated: (1)

The applicability to x of any tenn G 1 entails the applicability to x of any tenn G2 such that G2 is wider in extension than G 1.

In 28.1.0 the following principle is stated: (2)

The applicability to x of a general tenn G1 entails the inapplicability to x of any tenn G2 such that G 1 and G2 are both narrower with respect to a wider tenn G3 .

By stating the contrapositive of (1) we arrive at an equivalent statement: (1')

The inapplicability to x of a general tenn G 1 entails the inapplicability to x of any tenn G2 such that G2 is narrower than G 1 .

Statements (1 ') and (2) together yield the principle illustrated in the passage that reads "because particulars of another wider term are contradicted by the word 'sirhsapa" s own wider term": (3)

The applicability to x of a general tenn G 1 entails the inapplicability to x of a tenn G2 that is narrower than a tenn G3 that is incompatible with a term G4 that is entailed by G 1.

Statement (3) seems to be undeniable, for it apparently says no more than this: if two sets A and B have no members in common, then no subset of A has any members in common with a subset of B. Now although there are interpretations of (3) that confonn to our intuitions, other seemingly legitimate interpretations pose problems. An example of an interpretation that confonns to our intuitions is that if "Dachshund" is applicable to x, then "dog" is applicable to x; and if "dog" is applicable to x, then "bird" is not applicable to x; and if "bird" is not applicable to x, then "sparrow" is not applicable to x. An example of an interpretation that poses a problem is that if "oak" is applicable to x, then "tree" is applicable to x; and if "tree" is applicable to x, then "substance" is applicable to x; but if "substance" is applicable to x, then "quality" is not applicable to x; and if "quality" is not applicable to x, then "green" is not applicable to x. Thus, according to (3), it should not be possible to say "This is a green oak." Stated more generally, the problem is this: given any two general tenns G 1 and G2 , if a tenn G3 can be found that is wider in extension than both G 1 and G2 , then by principle (2) it would follow that G 1 and G2 cannot be applicable to the same thing at the same time. But this consequence if far too wide, for it rules out applying any word that names a particular quality to any object to which a word naming a substance is applicable, and so it becomes necessary to devise stipulations to modify the above rules in such a way that utterances that intuitively strike us as ill-formed (for example, "This Dachshund is a sparrow") are ruled out, while utterances that strike us a well-fonned (for example,

Chapter Seven

PramalJasamuccaya 5:29

293

"This oak is green") are not eliminated. To this problem the next several passages are dedicated.

29.0.0 The one term does not preclude the other directly, because the narrower term is not synonymous with the wider term. If this were the case, then it would not preclude other narrower terms. 29.1.0 But in this case "the one term does not preclude the other directly." Why? "Because the narrower term is not synonymous with the wider term." If "sirhsapa" excluded the jug directly, then it would be identical in meaning with "tree." "If this were the case,then it would not preclude other narrower terms." Just as the word "tree" does not preclude the pallisa and other kinds of tree, so also "sirhsapa," since it would be synonymous with "tree," would not preclude the palMa. Now one might think that this fault does not occur, because the wider term "tree" and the narrower term "sirhsapa" differ in that the former precludes fewer objects while the latter excludes more. But it is not the case that the reason that the two words express quite different objects is that "tree" expresses entities qualified by the tree-universal and "sirhsapa" expresses entities qualified by the sirhsapli-universal. But we do agree that the meanings of each word must be discrete. The wider and narrower terms preclude fewer and more objects respectively, but this preclusion is accomplished not directly by means of the words themselves but rather indirectly through the objects to which they apply. 29.1.1 This passage consisting of an objection and reply is unusually cryptic in Diilnaga's text, and the issue under discussion is far from self-evident. My translation reflects the extensive comments by Iinendrabuddhi (p305a8-306b 1). According to linendrabuddhi, the question is asked from the point of view of an opponent who is suggesting that a wider term such as "tree" and a narrower term such as "simsapa" have different extensions precisely because they have different intensions, and the extension of the narrower term is a subset of the extension of the wider term because the narrower term's intension includes the intensions of all terms that are wider than itself. On this account, the intension of "tree" is treehood, whiie the intension of "simsapa" is both sirtisapahood and treehood. Since simsapa has treehood as part of its intension, it directly excludes any object that does not have the universal treehood. That is, the word "simsapa" itself precludes any object that is not a tree. In contrast to this view, Iinendrabuddhi represents DiIi.naga's position to be that the narrower term "simsapa does not include treehood as part of its intension. Rather, the intension of "simsapa" is only the universal siIi1Sapahood, and the word itself excludes only those objects that are not simsapas. But since this property of being a simsapa, which is the sale property that the word "simsapa" makes known, occurs only in those loci in which the property of being a tree occurs, the property of being a simsapii is capable of functioning as an inferential sign by means of which the property of being a tree is known. Furthermore, since the

294

PramalJasamucca:ya 5:29

On linguistic signs

property of being a tree occurs only in loci in which the property of being a jug is absent, the property of being a tree is capable of functioning as an inferential sign by means of which the absence of the property of being a jug is known. In other words, "sirizsapii" does not itself exclude the property of being a pot, but it does make known a property which in turn provides the basis by which the preclusion of being a pot can be known inferentially. In order to see the implications of the distinction between Diimaga's view and the view against which he is arguing, let us consider what is involved in saying that a given word is applicable by convention to a given set of objects. To say that a term G is applicable to an object x is to say that x has the property that is the intension of G. For example, to say that "man" is applicable to x is to say that x has manhood, or, more simply, that x is a man. Thus we can write: (1)

"man" is applicable to x iff x is a man.

Let us now take into consideration the commonly acknowledged fact that all men are mortal, from which we can derive (2)

If x is a man, then x is mortal. From (1) and (2) it follows that

(3)

If "man" is applicable to x, then x is mortal.

The convention of the use of the word "mortal" is expressible in a way parallel to the convention of the word "man": (4)

"mortal" is applicable to x iff-x is mortal. And from (3) and (4) it follows that

(5)

If "man" is applicable to x, then "mortal" is applicable to x.

From statement (5) one might be tempted to conclude that the word "man" contains as part of its meaning (its intension) the meaning of the word "mortal." That is, one might conclude that "man" applies by linguistic convention only to mortal things. According to linendrabuddhi's interpretation, it is precisely this conclusion that Diimaga wishes to avoid. In his view, "man" is applicable by linguistic convention only to things that are men; it just happens to be a fact of the world that all men are mortal, and consequently it just happens to be the case that "man" is applicable only to mortals. But statement (5) is true not by dint of linguistic convention but rather by dint of a fact of the world. Thus the gist of 29.1.0 is that "simsapii" precludes the jug and so forth extensionally (arthatas) and accidentally (shugs kyis; *siimiirthyiit) but not intensionally (sabdatas) and necessarily (dngos su; siik:jiit).

Chapter Seven

Pramtif;asamuccaya 5:30

295

30.0.0 Since an adjective that is co-referential with another adjective is applicable to the same subject, there is no incompatibility with the substratum. 30.1.0 Given that a narrower term excludes the objects denoted by other narrower terms, how is it that an adjective such as "sweet," "sticky," "cool" or "heavy" is co-referential with another adjective? In this case, there is no incompatibility, because "since an adjective that is co-referential with another adjective is applicable to the same subject, there is no incompatibility with the substratum." Since stickiness and other qualities named by adjectives mingle in the same substance, the adjective naming the other properties are suitable to the sweet-tasting substance, so there is no problem in applying several adjectives to the same subject.

31.0.0 An alternative explanation is that preclusion is due to non-observation. A wider term would exclude what is narrower than itself. That is not the case, because one does observe that which is connected with other thingS.67 31.1.0 "An alternative explanation [for why one narrower term precludes the objects denoted by other narrower terms] is that preclusion is due to non-observation [of one term's application to members of the other's extension]." The alternative explanation is this: since one narrower term is not observed to apply to other particulars, the narrower term precludes the others. Some wrong-headed people believe that this non-observation is because of hostility to the objects named by the verbal symbols. We reject this view. But, one might argue, if this alternative explanation is correct, then "a wider term would exclude what is narrower than itself." If exclusion is due. to non-observation, then since a wider term is not observed [to express] its own particulars, it follows that it must exclude them. "That is not the case, because one does observe [a wider term applied to] that which is connected with other things." We observe that the wider term makes its particulars known through that for which it stands and through the context in which it is uttered and so forth. 31.1.1 Iinendrabuddhi (J308b6 ff) cites several examples of situations in which one is understood despite the fact that one uses verbal symbols that are insufficiently specific. One example will suffice here. A man sits down to eat and asks for salt. Despite the fact that "salt" is a very wide term that applies to many varieties of natural salts, one knows in this situation to bring the man just one variety of salt, namely, table salt. Thus a wide term here not only applies to but also expresses its own particular, and it does so by virtue of being uttered in a given context (prakaralJll). The purpose of this argument is to ~how that since wider terms do

296

Pramcil;asamuccaya 5:31

On linguistic signs

sometimes express their subextensions--although usually they cannot do so, and never can do so without the aid of context and other extralinguistic factors--they are not included within the scope of the rule "If WI never expresses that which W2 expresses, then WI precludes that which is expressed by W2." For clearly if W2 were a narrower term with respect to WI' then by the principle that wider terms cannot express their subextensions it would follow that application of WI would preclude application of W2 . This discussion is continued in the passage consisting of the last foot of verse 31 and all of verse 32 and its autocommentary. Since this passage is rather puzzling, however, I have decided to provide no English translation. Kanakavarman and Vasudhararak~ita differ in their interpretations of this passage, and Jinendrabuddhi deviates from his usual phrase-by-phrase commentarial style and offers instead a paraphrase of the general issue. All of these are good signs that DiIi.naga's original text was enigmatic at this point. The general issue, however, seems clear enough. Since general terms play a role in yielding information of their subextensions, although such information is more specific than the general terms themselves without contextual clues can yield, it cannot be said absolutely that a wider term fails to make its subextensions known. And therefore a wider term, unlike a contrary or contradictory term, does not exclude its own narrower terms. This is the case despite the fact that a wider term unaided by contextual clues cannot make its subextension known and thus prima facie seems to fall under the rule that whatever term does not express a given object excludes that object.

***** 33.0.0 At the level of verbal symbols, expression of a universal is the term's preclusion of contrary terms. There can be uncertainty arising from one and the same verbal symbol as to what its meaning is. Uncertainty arises also from several verbal symbols. 33.1.0 What a universal is at the level of objects is an object's exclusion of contrary objects. Similarly, "at the level of verbal symbols, expression of a universal is the term's preclusion of contrary terms." Just as the property of being created makes the property of being transitory known because it excludes the property of not being produced, a universal is expressed through a term that precludes contrary. terms, and, by means of that universal, the term makes its object known. But even in that case "there can be uncertainty arising from one and the same verbal symbol as to what its meaning is." From a verbal symbol such as "ak~a" uncertainty arises as to whether its meaning is part of a cart or something else. In this case it is a single verbal symbol that is ambiguous. "Uncertainty arises also from several verbal symbols [that all have the same form]. " In case doubt arises from a form such as "bhavati" as to whether it means "he becomes" or something else, although the verbal symbols are similar in form, the hearer should pay attention to the particular verbal symbol rather than the form it

Chapter Seven

PramalJasamuccaya 5:33

297

has in cornmon with another verbal symbol, just as one does in the expression "ka iha." 33.1.1 The Sanskrit word "a~a" can be applied either to the axle of a cart or to gaming dice or to several species of plant whose seeds were used to make gaming dice. If a lexeme is defined as a particular configuration of sounds to which a specific meaning is attached, then what we have here are severallexemes that all happen to have the same phonetic configuration, namely, the sound lakshal. In the case of the sound Ibhavatil, on the other harid, there is not one phonetic configuration representing several different lexemes, but rather there are two words that happen to have the same form. In this context a word (pada) is to be understood as a lexeme to which a verbal conjugational suffix (TIil) or a nominal derivational suffix (Sup) has been added, thus making it suitable to be inserted into a slot in a sentential schema to form a grammatically sound sentence. BHAVATI may be the outcome of affixing the finite verbal ending TI to the verbal root BHU, or it may be the outcome of affixing -the sixth nominal case marker I to the noun BHAVANT. Since the word's place in a sentence and other clues should make it clear which of several homomorphs one is dealing with, there should not really be any occasion for confusion in an accidental formal identity between two words. For example, despite the fact that the two sentences "kab iha" (Who is here?) and "ke iha" (Who are here?) have, through the application of external sandhi rules, the same pronunciation in rapid speech, when one hears the sound Ika ihal one understands from context which sentence is intended and ignores its homomorphism with the sentence that is not intended. Similarly, when one hears the sound Ibhavatil as part of a sentence, one understands which word is intended and ignores its homomorphism with the other word.

34.1.0 If a term expresses its own meaning by precluding other meanings, why do the faults mentioned above not arise? 34.1.1 In the next three verses Diilnaga attempts to show that the hypothesis that a verbal symbol expresses preclusion (apoha) is immune from all the arguments used in Prarniil:zasarnuccaya 5:2-11 to show that a verbal symbol cannot express particulars, universals, relations or instantiations.

34.0.0 Because a term is not observed to apply to objects in the extension of a contrary term, and because it is observed to apply to members of its own extension, it is easy to connect [the term to its meaning], and the term is not errant in its meaning.68 34.2.0 Association and dissociation are the two ways that a verbal symbol expresses its object. They consist respectively in applying to what is similar and in not applying to what is dissimilar. It is not necessary to say that' a verbal symbol applies to every instance of what is similar, because in some cases it is not possible to express an extension that is unlimited. But it is possible to say that it does not occur in the dissimilar--although it too is unlimited--simply on the basis of its not being observed to apply to

298

Pramal}asamuccaya 5:34

On linguistic signs

any dissimilar instance. For this reason, because a term is not observed to apply to anything other than that to which it is related, its expression of its own object is said to be a negative inference. If this inference were based on association, then the verbal' symbol "tree" would generate no uncertainty with respect to a thing as to whether it is being applied to a "simsapa" or some other kind of tree, but uncertainty would arise as to whether it is being applied to the property of being earthen, its property of being a substance and so on. But since the verbal symbol "tree" is not observed to apply to what is not composed of the earth element, the inference [from the application of "tree" to a subject to that subject's possession of the property of being earthen] is an inference by means of dissociation only. The following verse summarizes the above points: 34.2.1 As Jinendrabuddhi (p311a1-b 8) points out. Diimaga is here emphasizing once again the parallelism between verbal symbols and inferential signs, that is, properties whose presence in a locus indicate or yield information of other properties in that locus. Ditmaga's account of the minimum requirements for an inferential sign to yield lrnowledge of an inferable property is at Pramiir;asamuccaya 2:5cd, where it is stated that the inferential sign must occur in some loci other than the subject of inference but similar to the subject with respect to possession of the inferable property. This condition of a property's association with a second property is called anvaya, which refers to presence of the sign in at least some of the members of the subjectlike class. But a further requirement that an inferential sign must meet is its non-occurrence in loci that are not similar to the subject of inference with respect to possession of the inferable property. This condition of dissociation with a second property is called vyatireka. In order that a verbal symbol, when applied to a given subject of discourse, can indicate that subject's possession of a given property M, the verbal symbol must meet requirements similar to those that must be met by an inferential sign. The word must apply to some objects that are similar to the subject of discourse with respect to possession of M, which condition is association, and the word must not be applied to objects that lack M, which condition is called dissociation. If it were the case that one could, on the sale basis of having heard the term G applied to x, lrnow that whenever G is applied it is applied to x, then one could, on the basis .of hearing the word "tree" applied to an oak, lrnow that whenever one heard the word "tree" it was referring to an oak. On the other hand, one could not be certain on hearing the word a second time whether or not it was being applied to a thing having the property of being earthen, to a thing having the property of being a substance and so on. All these properties were present when "tree" was applied to an oak, and the problem would arise of how to determine which of these properties "tree" means. Does it mean all of them? Or only some? In actual practice, however, when one hears "tree" one is uncertain as to whether it is being applied to an oak or an elm, and one is certain that it is being applied to an earthen substance. Therefore, it is not the case that one can, on the sale basis of hearing G applied to x one or more times, lrnow that G always applies to x. One can lrnow that G always applies to x only by lrnowing that G is not applied in the absence of x.

35.0.0 The property of being a tree, the property" of being earthen, the property of being a substance, the property of being

Chapter Seven

PramalJ.asamuccaya 5:35

299

real, and the property of being an object of awareness are respectively the causes of four, three, two and one certainties. Taken in the order opposite the one given above, they are the cause respectively of four, three, two and one uncertainties~69 35.1.0 Neither a linguistic sign nor an inferential sign can make its object known except through a relation, because it is impossible to be aware of an object in its entirety when it has a multiplicity of properties. And since they do not express their particulars, they are restricted to their objects. Thus the two problems mentioned first (at 2.1.0) do not occur in our theory. Nor do the problems discussed next (at 2.3.0) occur in our theory. Why?

36.0.0 Since a tenn's denial of other tenns pervades narrower tenns, the wider tenn is not different from tenns narrower than itself. Because denial of others applies directly, and because it has no internal divisions, the properties of a universal are established. 36.1.0 "Since a tenn's denial of other terms pervades narrower terms, the wider term is not different [in its case markers] from tenns narrower than itself." A wider term whose function is to exclude other objects can be in grammatical agreement with terms narrower than itself, because it does not exclude its own particulars. Therefore, the problem does not arise of the wider term needing to be articulated with a case ending different from that of its narrower terms. The wider term applies in grammatical agreement with another thing that is expressed by both verbal symbols. Thus it is restricted to its own extension, because it does not apply by itself to any extension other than its own. Nor do the problems discussed last (in 4.0.0) occur in our theory. Why? "Because denial of others applies directly, and because it has no internal divisions." It is not the case that a term applies to its own particulars by means of including other meanings. Therefore, the fault of not entailing its own particulars because it is subordinate and the fault of being used figuratively do not occur [in a linguistic sign of which the meaning is construed as preclusion of contrary terms]. Moreover, since the mere preclusion of contrary meanings has no subdivisions, the problem of not being capable of expressing its particulars owing to their being unlimited in number does not arise. And since the exclusion of contrary meanings is not a thing, the problem of not being general owing to its failure to pervade does not arise. Thus, because all the aforementioned problems do not arise, it is correct that the only meaning of a linguistic

300

Pramiil;asamucca:ya 5:36

On linguistic signs

sign is the preclusion of contrary signs. And because of 'this, "the properties of a universal are established. ,,70 Only under our hypothesis are the properties of a universal established, namely, unity, permanence and the condition of belonging to each member of a class. Preclusion of contrary meanings has unity, because it is undivided. It has permanence, because its substratum is not destroyed. And it has the condition of belonging to each member in a class, because it is cognized in every object in the verbal symbol's extension.71 Therefore, since it is free from problems and possessed of excellent virtues, it is only things that are qualified by the exclusion of contrary things that a verbal symbol expresses.72

-- Notes--

Chapter 7. PramiilJasamuccaya V:" On the nature of signs in language" 1. The original Sanskrit for this verse is quoted by Kamalas-l1a in Santarak~ita 1968 ed. p. 539:

na pramanantaram sabdam anumanat tatha hi tat krtakatvadivat svartham anyapohena bha$ate. 2. Hattori 1968 tr.p. 177.

rnngon sum dang ni rjes su dpag tshad rna dag ni gnyis kho na stel gang gi phyir

rntshan nyid gnyis gzhal bya rang dang spyi'i mtshan nyid dag las gzhan pa'i gzhal bar bya ba med dol Irang gi mtshan nyid kyi yul can ni mngon su yin lal spyi'i mtshan nyid kyi yul can ni rjes su dpag pa'o zhes shes par bya'oll In all quotations of Tibetan from Hattori's editions, I shall, for the sake of uniformity with other Tibetan quotations in this book, use the Wylie system of romanizaqon rather than the system that Hattori uses.

Chapter Seven

301

3. Diimaga 1982 ed., p. 150. The Tibetan translation of the name Kapila is Ser skya pa. 4. SiiIhkhya Klirikii 6cd: "tasmiid api casiddham parok~am aptiigamiit siddham." 5. Isvaralq~l)a 1967 ed., p. 45. 6. Bhartrhari 1965 ed., p. 30, klirikii 30: na cagamiid fte dharmas tarkena vyavati~thate api yajjiiiinaIh tad apy iigamapurvakam

r~II)am

7. Bhartrhari 1965 ed., p. 31, klirika 36: praty~am anumiinaIh ca vyatikramya vyavasthitiil). pitrrak~aIJpisaciiniirt1 karmajii eva siddayaIJ

8. Santarak~ita 1968 ed., p. 539: "anumiine 'ntarbhiiva i$\al)." 9. Siintarak~ita 1968 ed., p. 539-540: "sa va vivak$ii tatkiiryatvad vacanat pratIyate." 10. This verse has been much discussed in the commentarialliterature. The whole verse reads in Sanskrit as follows: na jatisabdo bhedanam iinantyavyabhicaratal) viicako yogajatyor vii bhediirthair aprthak§rutel) What various commentators have said of the verse will be outlined in the sections that follow. 11. Siintarak~ita 1968 ed., pp. 341-342. 12. Santar~ita, Tattvasangraha 871-873: talra sva.lalclal)aIh tiivan na sabdail) pratipiidyate saIlketavyavahariiptakalavyaptiviyogatal) vyaktyatmano 'nuyanty ete ria parasparariipatal) desakiilakriyiisaktipratibhiisadibhedatal) tasmat saIlketa~~ 'rtho vyavahiire na drsyate na cagrhltasaIlketo gamyate 'nya iva dhvanel) l3. See the articles nlpa and svanlpagrahalJa in Renou (1957) for a discussion of the history of the usage of this expression in the writings of the Sanskrit grammarians. 14. Bhartrhari 1963 ed., pp. 122-128. 15. The verse reads:

302

Notes

jfianam prayoktur bahyo 'rthaI:t svarupaIh ca pratIyate sabdair uccitais te~am sambandhaI:t samavasthitaI:t 16. For a more complete discussion of this point in the history of Indian philosophy of language see Matilal 1971:42-44. 17. This quotation, which is entirely missing in Kanakavarman's translation and goes unmentioned by Jinendrabuddhi in his subcommentary, has been identified by Hattori (1975) as Vakyapadfya 3.14.8: vibhaktibhedo niyamiid gUI,lagulJyabhidhayinoI:t samanadhikaralJyasya prasiddhir dravyaSabdayoI:t. Cardona (1967-68:327-331) has cited the verse and given a .full discussion ofits import. 18. For Piirthasarathimisra's account see Kumarila Bhatta 1898 ed., p. 595. For Uddyotakara's see 1916 ed., p. 320. 19. The Sanskrit for this passage reads as follows: jfianam prayoktur bahyo 'rthaI:t svarupaIh ca pratIyate sabdair uccitais te~aIh sambandhaI:t samavasthitaI:t1l111 pratipattir bhavaty arthe ji'iane va samsayaI:t kvacit svarupe$iipalabdhe~u vyabhicaro na vidyatell211 asyayam vacako vacya iti ~a$thya pratIyate yogaI:t sabdarthayos tattvam ity ato vyapadisyatell311 nabhidhanaIh svadharmelJa sambandhasyasti vacakam atyantaparatantratviid riipaIh nasyapadisyatell411 20. See especially Bhartrhari 1963 ed., pp. 122-128. 21. DharmakIrti 1968 ed., p. 4-5: vaktvyaparavi~ayo yo 'rtho buddhau prakasate pramalJyaIh tatra Sabdasya narthatattvanibandhanam.

22. For a clear discussion of lexemes and forms thereof as I am using the terms here, please see chapter two of Matthews 1974. An approximate idea of a lexeme is that it is the form of a word that one looks for in a dictionary, in which it is customary to disregard inflectional forms that words take when they actually appear in sentences. The term "lexeme," therefore includes the senses of the terms "dhiilu" (verb root) and "pri'itipadika" (nominal stem) as used by the Sanskrit grammarians to refer to words to which the verbal affixes (TIn) and nominal affixes (Sup) have not yet been added. Only when a verbal or nominal suffix is added to a 1exeme is it called a word (pada) fit to be used in a sentence (wikya).

23. The original Sanskrit for this first foot Qf the fourth verse is quoted by both Bha\(aputra Jayamisra (in Kumarila Bhaita 1946 ed., p. 60) and Parthasarathimisra (in Kumarila

Chapter Seven

303

Bhatta 1898 ed., p. 598), where it reads "tadvato nasvatantratvat." KamalaSIla (in Santaraksita 1968 ed., p. 382) also mentions this foot of verse, inserting two words from PramafJasamuccaya 5:2 to make it into a complete sentence: "tadvato na vacakal:J sabdo 'svatantratvat." It is on the basis of this expanded sentence in KamalasIla that I provide the phrases in square brackets in my translation. 24. The original Sanskrit for these two sentences is quoted by KamalasIla in Santar¢ita 1968 ed., p. 382: "sacchabdo jatisvariipopasarjanariJ dravyam aha, na sakSad iti tadgataghatadibhedanilksepad atadbhedatve samanadhikaraI)yabhava-... " KamalasIla quotes these words of Ditmaga in his commentary to Tattvasamgraha 971, which in tum is a verbatim quotation of Kumarila Bhatta's MrmamsaSlokavarttika apohavadasthanaka verse 120, in which Kumarila Bhatta attempts to confute DiIinaga's position by the same line of reasoning with which DiIinaga tried to refute the position under discussion here. Hence both KamalaSIla's commentary to Tattvasamgraha 971 and Bha\taputra Jayamisra's commentary to Mfmamsa§lokavarttika apohaviidasthanaka 120 are rich in paraphrases of the line of reasoning DiIinaga is employing here. 25. Tattvasamgraha verse 971:

athanyapohavad vastu vacyam ity abhidhlyate tatrapi paratantratvad vyapti sabdena durlabha. 26. This passage occurs in Kumarila Bhatta 1946 ed., p. 60: "atra bhiksur aha--jatisabdal) sadadis tadvisistarn eva dravyam abhidadhan na tadgatarn eva gha(3.divisesarfipam ilksipati. paratantro hy asau sattliIh nimittI1qtya dravye pravartate na tu svatantral:J. sarvaviseSaQavisistaffi dravyariJ vakti. tatas ca ghatades tena sacchabdenanaksepan nasti tadvacina samanadhikaraI)yam. na hy ekopadhivisiste dravye 'bhihite paropadhina sarnanadhikaraI)yam bhavati. yatha madhuraSabdena madhuryopadhiviSistarh khaQQadravyam abhidadham tadgamparasukladyanaksepat tena na sarnanadhikaraI)yarh yathaitan madhurarh svetarn iti tathaitenapi na bhavyarh san ghataI:J iti--iti." 27. This sentence from DiIinaga's autocommentary was quoted by Bhattaputra Jayamisra in Kumarila Bhatta 1946 ed., p. 61: "sacchabdo 'pi bhfitarthena svariiparh jatirh vaha. tatra pravrttas tadvaty upacaryam3.I)e (gaUI)aQ syat)." 28. Uddyotakara's paraphrase (1916 ed., p. 321), "yac ca yatra varttarnanam anyatropacaryate na tattasyabhidhayakarn ... ," comes very close in meaning, if not in exact wording, to DiIinaga's original statement. Jinendrabuddhi's reading (P275b5) supports the reading in Vasudharar¢ita in which we find "gang zhig gang la... ," which is a virtually literal translation of "yac ca yatra," rather than the translation by Kanakavarrnan in which we [md "gang la .... " 29. Uddyotakara (1916 ed., p. 321): "sacchabdena pradhanyena sattayam varttate. tatra vartamanas tadvaty upacaryate. " 30. "Qualitative resemblance, whether it arise from the transfer of a notion of from the influence of the quality, is absent from the instantiation." The two Tibetan translations show that Kanakavarrnlm and Vasudhararaksita differ in their interpretations of this sentence. Kanakavarrnan's, which is supported by Jinendrabuddhi (P276a5), might be rendered back into Sanskrit thus: na pratyayasamkrantigufJopakarabhyam tadvati gUlJ.asarupyam api sambhavati. The syntax in Vasudhararaksita's translation, on the other hand, is very near the syntax in Uddyotakara's paraphrase (1916 ed., p. 321): "tadvati ca

304

Notes

na gUlJasiirupyiit pratyayasarhkriinti/;t... gulJopariigiit." Essentially the difference in interpretation is that Kanakavarman and Jinendrabuddhi understand pratyayasarhkriinti and gUl}opakiira to be two separate causes of gUlJasiirupya, while Uddyotakara and Vasudhararaksita understand gUlJasiirupya and gUlJopakiira (=gulJopariiga) to be two separate causes of pratyayasarhkriinti. Unfortunately, we get no clues from Kumarila Bhatta and his commentators, who remain silent on this discussion. Muni Jambuvijaya, in his retranslation in Sanskrit (Mallavadin 1976 ed., p. 607), follows Kanakavarman's interpretation. 31. Uddyotakara (1916 ed., p. 321) treats the example of applying the word "king" or "master" to a servant as an example of transferring an idea on the basis ofa similarity in quality: "gUI'Jasiirupyiit pratyayasarhkriintil;., yathii sviimisabdasya blzrtye." 32. Such an example is cited by S"amkara 1980 ed., p. 110 in his commentary to Brahmasutra 1.1.6: "athocyetacetane 'pi pradhane bhavaty atmasabdal)., atmanal). sarvarthakaritvat, yatha rajiial). sarvarthakaril)i bhrtye bhavaty atrnaSabdo mamatma bhadrasena iti. pradhanam hi puru~asyatrnano bhogapavargau kurvad upakaroti, rajiia iva bhrtyal). samdhivigrahadi~u vartamanal:1. (Now it might be said that the word "self' is applied to the material nature, even though this latter is insentient, because the material nature does everything for the self. It is similar to a king's saying with respect to a servant who carries out all his tasks, "Bhadrasena is my self." For the material nature aids the essential man, which is the real self, getting enjoyment and liberation for it, in the same way that a servant serves the king in peace and war.)" 33. The original Sanskrit for the second line of this verse was quoted by Bha\taputra J ayamisra: "gul)opakaratadriipye prakar~aI). syad vina dhiya." 34. Kumarila Bhatta 1946 ed., pp. 65-66. "atra bhik~ul)oktam. kiiicit kutracit vartamanam upakaram asajad vartate. yatha l~ sphatike 'nuraiijanopakarel)a vartate. vartate cej jatir dravye tatha tasyanurajanadyupakaral). kartavyal:1. tatas ca jatibuddhaya vinaiva tadupakrtadravyamatrapratItya bhavitavyam. na hi lak~arakte sphatike grhyam3.l)e tadgrahal)am aparam apek~te. na caivam atrasti. tena nasty upakarako dravyavrttil:1 iti." 35. This verse is quoted in both Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 600) and Bhattaputra Jayamisra, who refer back to the arguments given to repudiate the view that general terms express particulars (bheda): tadvams tu bheda evestal:1 sa ca pilrvam niraIqtal).. " 36. The two Tibetan translations for the first part of the commentary to this stanza are at considerable variance, differing both in vocabulary ·and syntax. The English translation, which must be regarded as very tentative, is based primarily on the interpretation given by Jinendrabuddhi as outlined in section 7.2.1 37. See 2.1.0. 38. The original dictum, which is cited a number of times in the literature of the Sanskrit grammarians, reads "samasakrttaddhit~u sambandhabhidhanam anyatra Iiiohyabhinnarupavyabhicaritasambandhebhyal).." 39. Kumarila Bhatta 1946 ed., p. 62: "yac catra bhiksUl)a jatimattvamatram vacyam aSaiJkya vikalpitam. atha jatimattvam kim jatitadvatol:1 samoandhal). kim va

?hav~yatIty

Chapter Seven

305

siimanyariipam iti. evarh ca vikalpya--piirVoktajatisarnbandhabhidhanado~o yojanlyal)--ity uktvoktarn. tadvanmiitre 'pi sambandhalJ sattii veti viciiritam tad api samanam." 40. See Hattori 1968:85, note 1.28. See also Hayes 1983. 41. A more complete explanation ofthis exception appears in Hayes 1983:712-713; 42. Patafijali, Vyiikaral)amahiibhii$Ya under PiiJ)ini Siitra 4.1.63: pradurbhavaviniiSabhylirfi sattvasya yugapad guQaiI) lIllarvaliIigam bahvartharh tiirtI jatIm kavayo vidul). 43. Somewhat different accounts of this dictum are to be found in Bhattaputra JayamiSra, Helaraja, Kaiyata, Haradatta Misra and Jfiiinendra. For details of how their accounts differ and where they occur see Hayes 1983. 44. Kumiirila Bhatta 1898 ed., p. 601: tadvan artho ghatiidis cen na ghatiidisu vartate siimanyam arthal) sa katham .... 45. Kumarila Bhatta 1946 ed., p. 63. This citation reads instead of "ghatiidi~u (bum sogs)."

"patiidi~u

(snam bu sogs pa"

46. See 8.1, which is quoted again by Jinendrabuddhi (P282b2) in his commentary on this statement. 47. Butchvarov 1966:7-8; emphasis added. Butchvarov's account of the arguments used by European nominalists occurs on pp. 16-55 of his 1966 work. 48. Rangaswami Iyengar (1927) and Hattori (1975) have both cited the original Sanskrit for this phrases, which is found quoted by both Piirthasarathimisra (Kumiirila Bhatta 1898 ed., p. 601) and Bhattaputra Jayamisra (Kumarila Bhatta 1946 ed., p. 64.): yady ekatriisiratadivat 49. The original Sanskrit for kiirikii lIb, quoted by PiirthasarathimiSra (Kumiirila Bhatta 1898 ed., p. 602) and Bhattaputra Jayamisra (Kumiirila Bhatta 1946 ed., p. 64.), has been cited by Rangaswami Iyengar (1927) and Hattori (1975): naitaj jiiter ajiitital). 50. The orginaI Sanskrit for this foot of verse as quoted by Piirthasarathimisra (Kumarila Bhatta 1898 ed., p. 602) and Bhattaputra Jayamisra (Kumiirila Bhatta 1946 ed., p. 64) has been quoted by Rangaswami Iyengar (1927) and Hattori (1975): art'hiiksepe 'py anaikantal).

306

Notes

51. Frauwallner (1959b: 102) and Hattori cite slightly different versions of -the Sanskrit original of this verse, which is quoted by Bhattaputra Jayamisra (Kumarila Bhatta 1946 ed., p. 46): bahudhapy abhidheyasya na sabdlit sarvathli gatiJ:I svasambandhlinuriipyeI)a vyavacchedlirthakary asau. Frauwallner's citation reads "bahutve 'pi" and "-anurupyiit" for "bahudhiipy" and "anurupyelJa" respectively. Frauwallner also calls attention to the similarity between this verse and PramiilJasamuccaya 2: 17, which he tentatively reconstructs as follows: bahutve 'py arthadharmliI)lirtl na lingat sarvatha gatil:J yenlinubaddharil tasyanyavyavacchedena gamakam. 52. Frauwallner (1959b:102) has tentatively reconstructed this verse: anekadharma sabdo 'pi yenlirtham nativartate pratyayati tenaiva na sabdagUI)atvadibhii:J. 53. Frauwallner reconstructs PramiilJasamuccaya 2: 17 as folIows: tathangaril yena riipeI)a IingiI)aril nativartate tenaivlinekadharmapi gamayati na cetaraii:J. 54. This full verse, as Hattori (1975) has pointed out, is quoted twice by KamalaSlla (Slintarak~ita 1968 ed., p. 365 and p. 379): apohyabhedlid bhinnlirthal:J svarthbhedagatau jaQlil:l ekatrabhinnakaryatvlid vise~aI)avise~yatli. 55. Hattori (1975) points out that this passage is quoted by KamalaSIla (Slintarak~ita 1968 ed., p. 379): na hi tat kevalam nIlaril na ca kevalam utpalam samudliyabhidheyatvaL .. 56. Ruegg 1959:37.) 57. Ruegg cites Viikyapadlya l.74: pade na varI)li vidyante vel1)e~v avayava na va vakyat padlinlim atyantam praviveko na kascana. 58. For more on this notion of atomic sentences among the Indian logicians, see Matilal 1966:385 ff. 59. This observation is based on the commentary of Iinendrabuddhi (P293b2).

60. See P294a2•

Chapter Seven

307

61. The Sanskrit original for PramalJasamuccaya 25cd has been quoted by Parthasarathimisra (KumariIa Bhatta 1898 ed., p. 606) and Bhattaputra Jayamisra (Kumarila Bhatta 1946 ed., p. 70): anyatve 'pi na samanyabhedaparyayavacinam Muni Jambuvijaya (Mallavadin Ksamasramana 1976 ed., pp. 638-639) has translated the verses, Diimaga's prose commentary and selections from linedrabuddhi's commentary for PramalJasamuccaya 5:25cd-28 into Sanskrit from the Tibetan translation of Vasudhararaksita, which is unfortunately the less reliable of the two Tibetan translations now available to us. 62. The example is provided by Jinendrabuddhi. In Tibetan the sentence reads "rgyal po'i skyes bu gzugs mdzes pa can shog," which might be retranslated into Sanskrit as "surupo rajapuru~a agacchati." 63. This verse has been quoted in full in KamalaSIla's Tattvasamgrahapafijikii (S"anta1968 ed., p. 379).

rak~ita

tanmatrl'ikaIi!GaIJad bheda1J svasamanyena nojjhital) nopatta1) samsayotpattel) sarnye caikarthata tayol). The first stanza has also been quoted by Jayamisra (Kumarila Bhatta 1946 ed., p. 72.) and Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 608). 64. There is a passage in Jinendrabuddhi's subcommentary for which I would suggest a slightly different translation from that given by Muni Jambuvijaya. The Tibetan reads (P303b 4): de dag khyad par dang khyad par gzhi nyid 'dod pa yin. zhes pa shing ni shing sha pa la sogs la 'khrul pa rnam par gcod pa la bItos so. zhes pa khyad par dang khyad par du bya ba'i dngos po yod do. shing sha pa ni shing gi spyi la mi 'khrul zhing bltos pa rna yin no zhes pa med do. Muni Jambuvijaya translates it (Mallavadin Ksamasramana 1976 ed., p. 639):

na tayor i~ta(s tulya?) viseSaIJaviseSyateti vrksal) sirhsapadivyabhicaravyavacchedam apeksata ity asti viseSllI~aviSesyabhava1J, sirh~apa tu vrksasamanyavyabhicaram napeksata iti nasti. I would suggest the following translation instead: tayor viSesaQavisesyateti vrk~al) sirhSapadivyabhicara vyavacchedam apeksata ity asti viseSaIJavisesyabhava1J. sirhSapa tu vrksasarnanyavyabhicara (vyavacchedarh) napeksata iti (viseSaIJaviseSyabhava1J) nasti. 65. This stanza is quoted by Parthasarathimisra (Kllmarila Bhatta 1898 ed., p. 605) and BhaWlplltra Jayamisra (Kllmarila Bhatta 1946 ed., p. 69 and 71):

308

Notes bhedo bhedantarartham tu virodhitvad apohate.

66. The Indians at the time of Ditmaga recognized five primary elements: earth, fire, water, wind and ether. Very approximately earth, water and fire correspond to the three states of matter: solid, liquid and gas. Therefore since both trees and jugs are solid material objects, both are earthen. 67. Rangaswami Iyengar (1927) and Hattori (1975) have noted that the first three feet of this verse appear in both Bhattaputra Jayamisra (Kumarila Bhatta 1946 ed., p. 71) and Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 608): adntatvad vyudaso va samanyam syat svabhedanut nanyayuktasya cLsrtatvat.... 68. As Hattori (1975) has noted, this verse is quoted in full by KamalaSIIa (Santarak~ita 1968 ed., p. 378): acirster anYaSabdarthe svarthasyamse 'pi darsanat sruteJ:! sambandhasaukaryam na castivyabhicarita. 69. Rangaswami Iyengar (1927) and Hattori (1975) have both recorded that the original Sanskrit for this verse was quoted by Jayamisra (Kumarila Bhatta 1946 ed., p. 73) and Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 611): vrk~atvaparthivadravyasajjiieyaJ:i pratilomyataJ:! catustridvyekasamdehe nimittam niscaye 'nyatha.

Frauwallner (1959b: 103) also cites the verse arid notes its similarity to Pramar;.asamuccaya 2:14. 70. Rangaswami Iyengar (1927) and Hattori (1975) have shown that the original Sanskrit for the fourth foot of verse 36 was quoted by Jayamisra (Kumarila Bhatta 1946 ed., p. 74) and Parthasarathimisra (Kumarila Bhatta 1898 ed., p. 611): jatidharmavyavasthitiJ:!. 71. Hattori Masaaki (1975) cites a passage in KamalasIla (Santarak~ita 1968 ed., p. 389) that parallels this very closely without-actually quoting Ditmaga verbatim: sarvatrabhedad asrayasynucchedat 1qtsnarthaparisamaptes ca yathakramam jatidharma ekatvapratyekaparisamaptilak~al)a apoha evavatis\hante, tasmad gUl)otkarsad apy arthantarapoha eva sabdarthaJ:! sadhuJ:!. 72. Hattori (1975) notes that this passage occurs in DharmakTrti's Pramar;.avarttika (1960 ed., p. 62-63): "sabdo 'rthantaranivrttivisistan eva bhavan aha."

Chapter 8 Conclusions It was pointed out in section 1.3 of the first chapter of this book that various scholars have taken different positions with respect to the question of what Dmnaga's principal motivations were in writing his treatises on logic and epistemology. Scholars whose main familiarity with the Buddhist epistemologists was with the works of Dharmakfrti and Santarak~ita quite naturally concluded that the chief preoccupation of the Buddhist epistemologists was apologetic and polemic, their agenda being to show the correctness of Buddhist teachings and the falsity of all teachings that were incompatible with Buddhist teachings. Thus such scholars as S. Mookerjee and D.N. Shastri were inclined to characterize the Buddhist logicians as a whole as warriors in an ideological battle with the Brahmanical philosophers over a number of classical metaphysical issues such as the criteria of reality, the existence of God, the existence of universals, and the nature of potentiality. Questions of logic and epistemology were in this view ancillary to the more central metaphysical questions. In contrast to this view, A.K. Warder argued that the questions of the nature and limits of knowledge were of far more central interest to the Buddhist logicians both in the capacities as philosophers and in their roles specifically as Buddhists, pointing out that the Buddhist path was built upon the foundation of ascertaining truth--seeing things as they really are--and that the enterprise of ascertaining truth is one that presupposes an investigation of the very methods by which one sorts truth from error; therefore, argues Warder, the study of epistemology could be seen as the central business of a practising Buddhist. Generally speaking I am strongly inclined to support Warder's observations on this point. But Warder, like Mookerjee and Shastri before him, was still inclined to see the school of Buddhist epistemologists as almost monolithic and to assume that DiIinaga and Dharmakirti had essentially the same agenda, which was to arrive at some satisfactory method of deciding what is true and then to go on to decide what is tru~ and [mally to live in accordance with the truth.

309

310

Conclusions

Scholars who questioned the assumption that a philosophical uniformity could be found in the works of Dirinaga and the works of Dharmakmi were E. Frauwallner and R. Herzberger. Frauwallner did not make a particular point of rejecting the assumption of uniformity, but his whole approach to the study of Indian philosophy was critical, methodical and hi:;torically oriented, and he was free of the traditional assumptions that so dominated the thinking of many of the modern scholars from India; he was, therefore, free to discover significant differences even among thinkers who had been traditionally regarded as scholastic confreres. Following out more fully the advantages of being free of traditional assumptions about the history of Indian philosophy, R. Herzberger made a most explicit point of showing that DharmakIrti was not only different from Dirinaga but that he "washed way" many of the philosophical accomplishments of his predecessor. Both Frauwallner and R. Herzberger have helped to advance our understanding of the Buddhist epistemologists to the point where we can no longer speak of them as being anything like a uniform school of thought with a single agenda. From now on we must treat each Buddhist thinker on a case by case basis rather than seeing each philosopher as merely a spokesman for one or another school of Buddhism. Moreover, we must recognize that the Indian Buddhist intellectuals, while being dedicated in some sense to the Buddhist way of doing things, were also trained first as Brahmanical intellectuals and were therefore heavily influenced by general Indian and not merely by Buddhist modes of thinking. It could be said, taking R. Herzberger's observations a step further, that Dharmakmi not only washed away Dirinaga's philosophical accomplishments but also washed away much of the accomplishment of the Buddha as well. It is ironic that in his very attempt to secure the truth of the traditional teachings of Buddhism, to establish the authority of the Buddha himself as a teacher, and to defend established Buddhist doctrine as much as possible from the sharp-minded critical attacks of some nonBuddhist thinkers, DharmakIrti managed to violate the essentially openminded and critical spirit of many of his predecessors, putting in its place a dogmatic edifice that eventually very nearly imprisoned a number of the later Indian and Tibetan Buddhist traditions. At least one important strain of the classical Buddhist philosophical spirit was non-assertive and skeptical in tone. While for many of the later Buddhist epistemologists the task of the philosopher was to arrive at correct understanding, for the skeptical tradition of earlier Buddhism the task of the philosopher was more to remove all forms of false security that blind one to one's own unimportance and fragility. Among the most insidious forms of false security, according to the skeptic, is the confidence that arises from

Chapter Eight

311

believing that one understands. This undeserved self-confidence, seen as the source of arrogance and lack of humility, could also be seen as fundamentally incompatible with the pervasive spirit of compassion within Buddhism. Compassion, the ability to be responsive to the pains and tolerant of the weaknesses of others, is most easily cultivated by those who are acutely aware of their own pains, for if one does not know one's own pain then one does not know pain at all, and if one cannot tolerate one's own weaknesses then cine cannot tolerate weaknesses in others. That a message somewhat like the one just outlined was the central message of some passages of the Sutta Nipata and several other PaIi texts is beyond dispute. That such a message was also at the heart of Nagarjuna's teaching is, I think, arguable, and I have tried to make the case for his skepticism in chapter two. It is less obvious that Diimaga was a skeptic, for unlike NagaIjuna Dirinaga does not make an explicit point of saying that he has no opinions and nothing to assert. I believe it is significant, however, that Dirinaga, who was so well aware of the philosophy of Bhartrhari and in fact took over many of the great grammarian's views as his own, never tried to answer Bhartrhari's insightful observations on the fallibility of inference fu"1.d sensation. Dirinaga did, however, go to great lengths to argue that scripture cannot be regarded as any more authoritative than sensation and inference. Whereas Bhartrhari's argument had been that sensation and inference are fallible and therefore human beings must rely on scripture to provide them certainty in morality and cosmological matters, Dirinaga's argument was simply that scripture cannot be any less fallible than inference. The overall view that Dirinaga seemed to have was that sensation can provide us knowledge only of phenomena and never of the external objects that are presumed to bel the causes of phenomena; in this he followed Vasubandhu and other phenomenalists. Inference, being based on induction in which there is always a risk of error, can never provide any new knowledge beyond what sensation itself can provide. The canons of logic serve only to eliminate umeasonable opinions, never to construct new opinions; in this Dirinaga followed Bhartrhari. And scripture can provide us only with the raw data out of which to make some conjectures about the thoughts of the human beings who wrote the scriptures, but no scripture can provide us with an access to realities that are not already accessible to us through sensation or inference; in this Dirinaga followed virtually the entire Buddhist tradition before him. And in any event, the only realities to which we have any access at all are the realities of our own private experiences, about which we can know only that when analysed they are full of internal inconsistencies and can therefore not correspond to things as they are

312

Conclusions

independent of our experiences of them; in this DiIinaga' followed Nagarjuna. Coming to tenns intellectuaily and emotionally with the limits of our understanding is fully as important within the skeptical traditions of Buddhism as is coming to tenns intellectually and emotionally with the fact of death, the limit of our existence. Realizing and accepting our own finitude and insignificance, whether through .the traditional methods of gazing at corpses and seeing the inevitability of our own death or through an analysis that reduces all our fundamental intellectual structures--all the persons, atoms, universals, and causes around which all thinking revolves--to absurdity and leaves us no respite from these absurdities of our own thinking, is for the Buddhist nirval}.a itself. The quest for such a nirval}.a was, I contend, the principal motivation behind DiIinaga's philosophical presentations.

Appendix A Glossary of Sanskrit Terms The following glossary contains Sanskrit terms used by DiIinaga the occurrence of which in the Prarni'ilJasarnuccaya has been attested by citations in other works. Enclosed in parentheses following each Sanskrit ,~erm is the Tibetan translation of that term as found in Kanakavannan' s translation. In cases where a different translation of the term has been used by Vasudhararak~ita, his translation will be preceded by "V=". A Tibetan phrase preceded by "J=" indicates the rendering of the term by BIo gro brtan pa's translation of Jinendrabuddhi. Further information on the terms as dealt with in the present work is contained in square brackets after the main discussion of the entry. References to the occurrence of the term in the Prarni'ilJasarnuccaya indicate the chapter and verse. Words in quotation marks within the square brackets refer to the subject and author index of this work. (krtaka) of smaller units, are anitya. [2:7]

arhsa (cha shas) Same meaning as bheda, which see. [5:34]

anumiina (rjes su dpag pay

atadbhedatva (tha dad pa rna yin pa de nyid; V= de nyid tha dad pa nyid) The fact of not being a

Inference, jUdgement, reasoning; especially, the process of drawing conclusions concerning the presence or absence of certain properties in a locus on the basis of observing other properties in that locus. More generally, the term as used by Diilnaga signifies any cognitive act of which the subject matter is wholly or partly made up of universals, names or synthetic·judgements. [2:5,5: 1. See "inference" .]

proper subset of some specified extension. [5:4]

adrsti (rna rnthong) Non-observation; especially non-observation of a counterexample to a general rule. In Ditmaga's system, a rule is good so long as there is no observation of a counterexample. [5:31,34]

anitya (rni rtag) Transitory, subject to

anumeya (rjes su dpag bya) The

change, impermanent. All complex t/Iings, being composed or created

inferable object. The object about

313

314

Glossary of Sanskrit Terms

which the inference supplies information. [2:5]

anaikiinta (rna nges pa) Said of a property that occurs sometimes with and sometimes in the absence of a second property and therefore cannot be used as evidence for that second property. [5:11]

aprthakSruti (tha dad med thos; V= so sor rjod byed min) The fact of two or more words being used without a difference in case-endings. Several words appearing together in the same sentence with the same case endings is one of the formal indications of a samanadhikaralJ.ya relation between the words. [5:2]

apoddhiira (dmigs kyis bkar) Grammatical term designating the analytic process of reasoning by which one determines the meanings of individual words, morphological elements and other components of sentences by comparing how the meanings of sentences vary when components within those sentences are replaced by other components. [5:46]

apoha (sel ba) Most often occurs as part of the phrase anyapoha (gzhan sel baY. Separation, division, . discrimination, preclusion, exclusion; especially, a word's exclusion from its own domain of application of a) the applicability of contrary terms and b) the objects to which contrary terms are applicable. [5: 1. See "abstraction from others," "anyapoha," "apoha theory of meaning" in index.]

apohate (set bar byed) Finite verb form from which the abstract noun

apoha is derived. [5:28]

aprameya (gzha/ bya min) Unknowable. The property of being unknowable is a stock example of a property that can never serve as an inferential sign, since it can never be observed to occur in a subject, or in any object in the induction domain, and therefore does not satisfy any of the three characteristics of proper evidence. [2:7]

abhidhiina (brjod pa; V= mngon pa brjod) The action of naming or expressing. [5:9]

abhidheya (brjod bya) That which is communicated or made known through a verbal symbol. [5:12,15]

abhinnariipa (tha mi dad pa'i ngo bo; V= tha dadpa'i [sic] dngos po) Grammatical term for words having the same phonetic form whether they name universals or possessors of universals. Names of colours and some other adjectives are abhinnarllpa. In contrast, usually the name for a universal is derived from the name for a possesor of the universal by the addiction of an abstraction suffix. "purhsa" =male, and "purhsatva" = masculinity; but "sukla" = the colour white-in general, and "sukla" = a white thing in particular. [5:9. See "bhavapratyaya".]

amiirta (Ius min; = Ius can min) Incorporeal, not having a physical body. The property of incorporeality is Ditmaga' s stock example of a pr(\perty that has the first two characteristics but fails to have the third characteristic of proper evidence. Incorporeality occurs in sound, the stock subject of inference, and also occurs in such objects as action in the

Glossary of Sanskrit Tenns indllction domain that possess transitoriness, the property to be confinned. But since it is not always absent in the absence of the property of being transitory, for ether is not transitory despite being incorporeal, it cannot be used as evidence for transitoriness. [2:7]

artha (don) A thing, an object; especially, an object for which a symbol stands, a symbol's meaning.

[5:9]

arthiik$epil (don gyis 'phangs) Disjunctive syllogism; a particular kind of argument or line of reasoning in which it is first established generally of a set of alternatives that at least one alternative of the set is true, and then one specific alternative is affirmed by denying all the others. [5:11]

aviieya (brjod min; V = brjod par mi bya) Inexpressible. Objects of sensation are held to be inexpressible, since there is no symbol that can apply uniquely to a sense datum. Relations are also held to be inexpressible in sentences in the sense that there is no word in a sentence that names a relation directly; rather, relations are indicated by case endings and other morphological features. [2:10]

avisarhviida (mi slu bay Nondeceptive, not misleading. It is an essential feature of a criterion of knowledge that it not lead one astray. [2:5]

avyabhieiira (' khrul med pa, mi 'khrul pa) Inerrancy; the fact of a tenn or property being restricted to only that in which a given property is present. [2:8,11. 5:9]

315

asvaiantratva (rang dbang med) Lack of independence. As used by Diimaga, the tenn signifies the state of being grammatically subordinate. [5:4. See upasarjana.]

iikiilik$a1J.a (' dod pa) Anticipation. Grammatical tenn for the expectation created in the hearer of an expression for some further infonnation. For example, when one hears "John's ... ", one expects to hear the thought completed by the name of something that John owns or is related to in some other way. Moreover, ... [5:26]

iidhiira (rten) Container. Diimaga's example of a relatum that is assymetrically related to another. Containment is an assymetrical relation in that if A contains B, then B does not contain A. That which contains is called adhiira. while that which is contained is called iidheya (brten.) [2:10]

iinantya (mtha' yas; V= mtha' yod min; J= mtha' yas nyid) Innumerability; the state of being infinite in number or at least beyond the practical possibility of counting. Because the objects to which a symbol can be applied are beyond reckoning, Diimaga argues that nothing to which the symbol is applicable is made definitively known through the symbol. [5:2]

iipta (yid ehes) A credible person, an authoritative witness or expert whose testimony is not liable to be corrupted by personal prejudices or incompetence. It is axiomatic in Indian religious philosophy that the reliability of scripture is dependent upon and proportional to the reliability of those who transmit and interpret it; consequently, it was not

316

Glossary of Sanskrit Terms

uncommon for Buddhists and others who rejected the authority of scriptures to impugn the integrity of the brahman priests who transmitted them. [2:5]

iisraya (rten) A locus of a property; a substratum. [5:36]

upaciira (btags pa; V= nye bar brtags pay The use of a term in a non-literal sense; the application of a term that normally applies to a given object to some other closely related or similar object. Simile, metonymy, synecdoche, etc. [5:4]

upasarjana (gtso bo ma yin pa tsam; V = khyad par du byas pay Grammatical term signifying a word that loses its independent status and comes instead to qualify another term as a result of compounding or the acquisition of certain morphological elements. [5:4]

ekatva (geig nyid) Unity. [See jatidharmaJ.

ekiirthatii (don geigy The fact of two or more expressions applying to the same object or set of objects. [5:26]

krt (byed pa; J= byed Idan) Grammatical term applied to all suffixes that, when applied to verbal roots, yield non-fmite verbal forms or deverbative nouns. [5:9]

krtakatva (byas; V= byas pay The property of being created or produced. The stock example in DiIinaga's logic of a property whose presence in a locus serves as an inferential sign of a further property, transitoriness, in that locus. [2:7. 5: 1]

krtsniirthaparisamiipti (mtha' dag rtogs pa; V= don ma Ius pa rtogs pay Literally, complete throughout the entire object. Said of a universal's occurrence in the particulars in which it inheres. The entire universal is supposed to reside without divisions in each one of pluraJ.ity of particulars. Diimaga' s own gloss for pratyekaparisamapti, which see. [5:36. See also jatidharma.]

gUl}a (yon tan) A quality. [5:2,5,40] gUl}opakiira (yon tan gyis phan 'dogs; V= yon tan gyis phan btags) The influence on one substratum of a quality that belongs to another substratum, owing to which influence the fust substratum appears to be qualitatively similar to the second. A stock example is that of a rose placed next to a crystal so that the red colour that actually resides in the rose apparently resides in the crystal, making the crystal virtually red. [5:5] ciik~u~atva

(mig gi gzung bay The fact of being visible. The property of visibility is DiIinaga's stock example of a property that meets the second and third characteristics of legitimate evidence but fails to meet the first. Visibility occurs with such transitory objects as pottery in the induction domain, and no non-transitory thing is visible. Therefore, whatever is visible is transitory. But visibility does not reside in sound, so it cannot serve as evidence for the presence or absence of transitoriness or any other property in sound. [2:7]

jatfa (blun; V= rmongs) Dull, stupid. Dumb, mute. DiIinaga claims that

Glossary of Sanskrit Terms words are jacja with respect to the particulars to which they apply; that is, words can name only general features of things but not the features of an individual that uniquely occur in it. [5:14]

jiiti (rigs) A universal. [5:4,11,36] jiitidharma (rigs kyi chos) The three characteristics of a universal, namely, a) unity (ekatva), the property of having no internal divisions regardless how many instances it resides in or applies to, b) constancy (nityatva), the property of not being destroyed when individual instances perish, and c) residence in a plurality of substrata (pratyekaparisamapti). [5:36]

jiitisabda (rigs sgra) A general term, that is, a term that names a universal. One of the five types of word discussed by Difmaga. [5:4]

taddhita (de la phan pay Grammatical term for a secondary suffix, that is, a suffix that unlike a krt suffix (which see) is not added directly to verb roots (dhatu) but rather is added to nominal forms that are themselves produced through the addition of a krt suffix to a verb rooL [5:9] .

317

dissociation from the absence of the property to be inferred. [3:1]

dariana (mthong; V = ston pay Observation, direct experience. In Difmaga's system of epistemology, direct observation takes precedence over both general reasoning and authority in that the general rules of either can be overturned by what one actually experiences. [5:34]

dravya (rdzas) Substance. In Nyaya and Vaise$ika metaphysics, substance is that in which actions (karman) and qualities (gUI)a) inhere. [5:2,4,35]

dharma (c hos) A property or attribute in the widest senses of those terms; that which is named by the predicate term in a proposition. [2:8,9,11]

dharmin (chos can) The locus of a property; that which is named by the subject term in a proposition. [2:8,9,11]

dM (blo) Idea, mental image. [5:5] nitya (rtag pay Imperishable, constant, immutable. [2:7; 5:36]

nityatva (rtag pa nyid) Constancy, unchangeability, permanence. See

jiitidharma. [5:36]

tadvat (de ldan) Equivalent tojiitimat vastu, a particular object in its role as a concrete locus of a universal; an instantiation of a universal. [2: 10. 5:4,8,9]

tririipaliizga (tshul gsum pa'i rtags) Inferential mark that has three characteristics: 1) residence in the subject of the inference, 2) association with the presence of the property to be inferred, and 3)

nimitta (rgyu mtshan) In general, the occasion or warrant for or grounds or basis on which a thing or state of affairs comes into being or continues to be. Of particular importance in the context of grammar, nimitta in sabdapravrttinimitta is the ground or warrant for applying a given word to a given objecL For example, the presence of the quality (gUJ)a) of white colour in an object is the grounds

318

Glossary of Sanskrit Terms

(nimitta) of the application (pravrtti) of the adjectival word (gur;asabda) "white" to the object. Presence of action (kriya) is the grounds of applying a verb or deverbative (kriyasabOO), and so forth. [5:35]

nivrtti (ldog pa nyid; V= bzlog paY Same meaning as apoha, which see. [5:36]

niscaya (nges pay Certainty; the absence of doubt as to whether a cognition is accurate or a proposition is true. [5:35]

pada (tshig) An individual word as a component of a sentence. By the conventions of PaI)ini, a sentence-ready pada consists of a verb root (dhatu) plus a conjugational affix (Tin) or a nominal stem (pratipadika) plus a nominal affIx (Sup). [5:46]

paryiiya (rnam grangs) Synonymous expression. [5:26]

piirthiva (sa las gyur; V= sa las byung) Earthen; being composed of the earth element. prakar~a

('phel ba; V= 'jug pa; J= rab gyur) Vividness, intensity.

or being present in a plurality of substrata. [5:36]

pramiil}a (tshad rna) Knowledge. A source of know ledge. The process of acquiring knowledge. Criterion by which an opinion or hypothesis can be judged. [5:1,50]

pravrtti ('jug pa; V= rab tu 'jug) Same meaning as vrtti, which see. [5:50]

bhiivapratyaya (J= dngos po'i rkyen) Grammatical term for the abstraction or nominalizing suffixes TV A [--> -tva] and TAl [--> -tal According to PaI)ini Siltra 5.1.119, these two secondary suffixes are used in expressing a thing's nature bhava). [See "abstraction suffIxes".]

bheda (khyad par; V= tha dad) Part of a set of objects, a subset; especially, a part of a set of objects to which a given term is applicable, a subextension of that term; in some contexts the bheda signifies an individual considered as a member of a given set. [5:2,8] A term that names a subextension of a given term, a narrower term with respect to a given term. [5:25,26,28,31] A difference [5:14]

eminence. [5:5]

pratibhii (spobs pa; V= so sor snang; J= so sor spobs) An image in cognition, an idea; especially the idea that is the content of a cognitive act consisting in an understanding or grasping of a state of affairs, the communication of which idea is the function of a sentence. [5:46]

pratyekaparisamiipti (re re la yang su rdzogs pa; V= so so ba kun la khyab pa) The fact of residing

miirttatva (lus can) Corporeal, possessing a physical body. One of Diimaga's examples of a property that, failing to occur in the subject sound, cannot serve as evidence for the presence or absence of any other property in sound. [2:7]

riirJhi (bstan pa; V and J= grags pay Grammatical term for an idiomatic expression, one the meaning of which cannot be derived from its morphological components

Glossary of Sanskrit Terms but is derived only from common usage. [5:9]

lak$a1].a (mtshan nyid) A characteristic mark; an identifying property possession of which identifies an individual as a member of a given class or as the very individual that it is. [5:36]

Ziizga (rtags; V= gtan tshigs) A sign; especially, an inferential sign. A property the presence of which in a specified locus is capable of yielding knowledge of a further property in that locus. The property indicated by the inferential sign is called iiJigin (Tlags can), or sign-bearer. [2:8,9,10,20]

variate ('jug Finite verb form from which vrtti is derived, which see. [5:9] .

viikya tshig A statement [2:5] (ngag) Grammatical technical term signifying a complete sentence, especially in contrast to pada, an individual word. [5:46];

viicaka (rjod byed, brjod byed; V= sgra, brjod par byed) That which expresses; a verbal symbol that is capable of making a thing known. [5:2,50]

viicya Same meaning as abhidheya, which see. That which is expressed by

a vacaka.

vibhakti (V= rnam dbye) Grammatical term denoting case terminations of nouns lmd personal endings of [mite verbs, called sup and tiJi respectively. [5:2]

virodhitva (' gal ba) The fact of being incompatible. Enmity, hostility. [5:28]

319

viSe$a1].avise$yatii (khyad par khyadpar gzhi; V= khyadpar byed dang khyad par bya) The relationship between a term that qualifies (vie$aTJa) and a term that is qualified (viserya). The qualification relation whereby one term specifies a part of the extension of another. [5: 14]

vrtti ('jug; V= yin)) The occurrence of a property in a locus. The applicability of an expression or a symbol to an object. [2:20. See "applicability. "]

vyabhiciira (' khrul pa) Errancy. The fact of a symbol or a property being able to apply to or occur in that in which a given property is absent, in which case the symbol is said to be errant with respect to that property. The fact that one property's occurrence is not restricted to another property. [5:2,34]

vyavaccheda (rnam par bead pa) Same meaning as apoha. [5: 1]

vyudiisa (sel ba; J= sel ba nyid) Same meaning as apoha. [5:31]

sabda (sgra) Sound in general. Particularly, articulate sound, language. A linguistic item, such as a word, term, expression, or symbol.[5:2,12,34]

siibda (sgra las byung ba) Knowledge derived from hearing symbols spoken. The process of . acquiring knowledge through heanng speech. The spoken testimony of reliable authorities as a source of knowledge. [5:1]

320

Glossary of Sanskrit Terms

sriiva1)a (mnyang bya) Audible. The property of being audible is DiIinaga's stock example of a property that has the fIrst and third characteristics of legitimate evidence but fails to have the second. Being audible does occur in sound, it it does not occur in anything in the induction domain that does not have the property of being transitory. But it also does not occur with anything in the induction domain that does have the property of being transitory. Any property that, being unique to the subject of an inference, cannot be observed at all in the induction domain cannot serve as evidence for the presence or absence of any property in the subject. [2:7]

fjafj(hl (drug pa) Shortened form of ,a,{hfvibhakti, sixth (nominal) case ending. Grammatical term for the possessive case in nouns. [2: 10]

samsaya (the tshom) Uncertainty; especially, the cognitive state resulting from inconclusive evidence. The lack of conclusive criteria by which to decide between two contradictory propositions. [5:26]

sattii (yod pa; V= yod pa yin; J= yod pa nyid) Reality. Actuality. Presence. In Nyaya and Vaise~ika metaphysics, the most general universal. [5:9]

samdeha (the tshom) Same meaning as samsaya. [5:35]

samiisa (bsdu ba; V= tshig sdud) Grammatical term for nominal compounds wherein all words except the last lose their case terminations. [5:9]

samudiiya (bsdus pay A complex entity; an individual existing as a

collection of components or a multiplicity of properties. [5: 15]

sambandha (' breI pay Relation, relationship. [2:8,10,20.5:9,12,34]

sambandhasaukarya (' brel jJa sla ba; V= ' breI pa shin tu sla; J= , brel pa sla ba nyid) Ease or feasibility of making or understanding a relation, especially between a symbol and what the symbol stands for; such a relation must be understood before a symbol can yield knowledge of that for which it stands. The easier the connection between sign and signified is to make, the more efficiently the sign can signify. [5:35]

siikfjiit (dngos su) Directly or immediately. Explicitly. opposite in meaning to arrhat, arthena or arrhatas, which terms are used in the sense of implicitly. [5:4]

siimiiniidhikara1)ya (gzhi mthun pay Grammatical term signifying 1) the residence of a plurality of properties in the same locus, and 2) the application, warranted by such coresidence, of a plurality of words to the same object or set of objects. [5:2,4]

siimiinya (spyi) Applied to terms or properties in the sense of that which is shared, general or common. [2:5 {where V= mtshungs). 5:10,25,26.]

svarupa (rang gi ngo bo) The form of a word. The phonetic characteristics of a word token that identify it as belonging to a specified word class. [5:4]

Appendix B Tibetan-Sanskrit Lexicon The following list of Tibetan terms those for which a Sanskrit equivalent has been attested in the second or fifth chapters of the Pramiil;asamuccaya. Following the Sanskrit equivalent of each term is an indication of the passage of the Pramiil;asamuccaya in which its use is attested. Numbers preceding a colon refer to chapter numbers, and numbers after the colon to verse numbers; numbers in italic print indicate that the citation occurs in Dirinaga's prose commentary to the verse of that number rather than in the verse itself. Terms that are taken from Kanakavarman's translation are marked by a "K" before the chapter-verse information, and those taken from Vasudhararak~ita's translation are indicated by "V." In case both translators use the same Tibetan expression to render a Sanskrit term, only the chapter-verse citation is given. Terms taken from Blo brtan's translation of Jinendrabuddhi's Pramiil;asamuccayatikii are indicated by "J," followed by the folio number to the Peking edition of the Tibetan Tripitaka. The words are listed in the order of Tibetan lexicons.

KA

KHA

dkar po min asita (K5: 10)

kha cig kecit (2:8)

rkyang kevala (V5:15)

kha dog varJ.la (5:51)

bskyed pa utpatti (5:26) upajanyate (K5:46)

khona eva (J313b6)

322

Tibetan-Sanskrit Lexicon khyad par bheda (K5:2,8,14,25,26,28,31) khyad par khyad par gzhi yin vise~aJ)avise~yata (K5:14) khyad par can visi~ta (K5:51)

grub (pa) siddhatva (2: 8) prasiddha (2:9,11) 'gal ba virodhitva (5:28; J304a2) 'gyur syiit (5:5,31; 278a 8)

khyad par du byas pa upasarjana (V5:4) visi~ta (5:36)

'gyur gzhan anyathii (V5:35)

khyad par byed dang khyad par bya vise~aJ)avise~yata (V5:14)

'gro rcchati (K5:40)

khyad par byas visi~ta (V5:41)

rgyu rntshan nirnitta (5:35; J312a 3)

'khrulpa vyabhicara (5:2)

sgru rna miiyii(5:40)

'khrul pa nyid vyabhicarita (5:34)

sgra dhvani (2:7) viicaka (V5:2) . -viicin (K2:25) sabda (5:2,12,34,50; BlObB) sruti (B11a 1)

'khrul rned pa avyabhicarita (5:9) avyabhicara (K2:8)

GA go bar bya ba yin garni~yati (V2: 11)

sgra la byung ba siibda (5:1)

NGA

go rim log pa las priitilomyatas (K5:35)

ngag viikya (5:46)

gyur rcchati (V5:41)

nges (pa) niscaya (5:35; J312a 3) niyarna (V5:2)

gyur pa priipta (5:41) grags pa riiqhi (V5:9; J281b2) grangs sarilkhyii (5:50)

dngos tadriipya (V5:5) dngos pa'i rkyen bhiivapratyaya (J281a7)

323

Tibetan-Sanskrit Lexicon

324 dngossu sak~at

(5:4)

mngon par brjod abhidhana (V5:9) sngo asita (V5: 10; J283a6) sngon po nna (5: 15; J291a 8)

CA ci ste

atha (K2:9) ci phyir

kimartham (2:9) ci zhig

kim (2:9) gcig nyid ekatva (5:36) gcig la ekatra (5:10,14)

CHA cha shas arnsa (5:34; J3lOb~ avayava (V5:50) chu salila (V5:50) chu tshogs salila (K5:5 0) chos dharma (2:8,9,11) chos can dharmin (2:8,9,11) mchog parama (5:40)

JA ji Ita ba katham (V5:1O) ji ltar katham (K5:1O) 'jug vrtti (K2:20) 'jug pa prakar~a (V5:5) pravrtta (5:4) prayujyate (5:50) pra vrtti (K5:5 0)

'jug min na vartate (V5:9) 'jug par byed pravartate (K5:50) rjes (su) mthun las anuriipyel)a (K5: 12; J289b 6) rjes (su) dpag Cpa) anumana (2:5; 5:1) rjes su dpag bya anumeya (2:5) rjes su dpog byed anumeyata (K2:9) rjod (par) byed vacaka (K5:2; V5:50) brjod (pa yin) abhidhana (K5:9) abhidhayin (V5:2) -akhya (V5:46) aha (5:4) i~(a (5:8) -vacin (V5:25) brjod (par) bya (ba) abhidheya (5:12,15; J288b6)

Tibetan-Sanskrit Lexicon brjod (par) byed bhasate (K5:1) vacaka (K5:50)

rtogs . gati (5: 12,14; J289b')

rtogs par 'gyur brjod par my bya

gamisyati (K2:11)

avacya (V2: 10)

stabs gcig pa brjod min

sahabhii (V2:20)

avacya (K2: 10)

ston par byed disyate (V2: 11)

NYA nye bar bskyed

stong tucchaka (V5:40)

upajanyate (V5:46)

ston pa nye bar brtags pa yin

darsana (V5:34)

upacaryamaQa (V5:4)

ston par byed nyid eva (5:8)

gnyis po dvayam (2:10)

mnyang bya sravaQa (2:7)

bhasate (V5:1) dyotayati (V5:1)

bItos pa med pa nirapeksa (5:50)

brten adheya (2:20)

bstan pa rii<;lhi (K5 :9)

TA gtan tshigs Iiilga (V2:8,9,1O)

bstan par bya disyate (K2: 11)

btags pa yin upacaryamaQa (K5:4)

rtag pa nitya (2:7)

THA tha dad bheda (V5:2,8,14,25,26,28,31) bhinna (K5:14)

rtag pa nyid nityatva(5:36)

tha dad pa rna yin pa de nyid atadbhedatva (K5:4)

rtags Iiilga (2:20; K2:8,9,1O)

tha dad med abhinna (5:14)

rtags can Iiilgin (2:20)

tha dad med thos aprtbakSruti (K5 :2; J272b 1)

rten adhara (2:20) asraya (5:36)

tha mi dad pa abheda (5:36)

325

326

Tibetan-Sanskrit Lexicon

tha mi dad pa'i ngo bo abhinnarupa (K5:9)

de dang Idan pa tadyuktam (2:11)

thams ead sarvatha (K5:12)

de Idan tadvat (2:10; 5:4,4,8,9; J281a6 , 313a8 , 313b6)

thigs pa bindu (5:50)

de la phan pa taddhita (5 :9)

the tshom samdeha (5:35) samsaya (5:26)

der gtogs pa tadgata (K5:4)

thos

don srilyeta (K2: 10)

artha (5:9,46)

mtha' dag rtogs pa lqtsnarthaparisamapti (K5:36)

don gyis 'phangs arthalqepa (5:11)

mtha' yas anantya (K5:2)

don gyis 'pheng arthak~epa (J284a4)

mtha' yod min anantya (V5:2)

don gcig ekarthata (5:26)

mthun tulya (V2:5)

don gcig nyid ekarthata (B03a8)

mthong darsana (K5:34; BlOb~

don phye bhinnartha (V5:14)

mthong ba dr~ta (5:31)

don rna Ius pa rtogs pa lqtsnarthaparisamapti (5:36)

dr~ti

(k5:40)

don rnam geod vyavacchedarthakarin (V5: 12)

DA dang atha (2:5) dang po Mau (5:46) de ngo bo tadriipya (K5 :5)

don la rnam beod byed vyavacchedarthakarin (K5: 12) drugpa $ll$thI (2: 10) 'dus pa samudaya (V5:50) 'dod (pa)

de nyid tha dad pa nyid atadbhedatva (V5:4) de !tar atha (V2:9)

a.Jcank~a (5:26; J303a3) icchanti (2:8)

Tibetan-Sanskrit Lexicon

PA

ldan yuktil (K5:31) yoga (5;2)

dpag mlyate (V2:9)

ldogpa nyid nivrtti (K5;36)

dpag par bya ba meya (V2:8)

bsdu ba samasa (K5;9)

dpag bya anumeyata (V2:9)

bsdus pa samudaya (5;15)

dpog meya (K2:8) dpog par byed mlyate (K2:9)

NA gnas -stha (2;20)

spangs pa yin niralqta (K5:8)

rnarnkun sarvatha (V5:12)

spong (rni) byed (na) uiihita (K5:26)

rnarn grangs paryaya (5:26)

spobs pa pratibha (K5:46)

marn par bead pa vyavaccheda (5:1)

spyi

rnarn par brtags vivecita (5:46) rnarn par gnas avati~thante

(5:36) vyavasthiti (5:36)

samanya (K2:5; 5:10,25,26)

PHA 'pheJ ba utkar~a (K5:36) prakar~ (K5:5)

marn pa dphyad vicarita (5:9) marn par rna chad pa anuccheda (V5:36)

BA burnpa ghata (5:4,9)

rnarn dbye vibhakti (V5:2)

bor (rna) yin (na) uiihita (V5:26)

snangba

byas (pa) !qtakatva (2:7; 5;1)

dr~ti

(V5:40)

snarn bu pa(a (V5:9)

327

328

Tibetan-Sanskrit Lexicon

byed Idan

rna bzung

lqt (J281a8)

byed pa

nopatta (5:26)

rnangpo

lqt (5:9)

blangs (rna) yin

bahudha (5:12)

rni 'khrul pa

(na) upatta (J3037)

bIun

avyabhicara (K2: 11)

rni 'khrul pa nyid

jal;ia (K5:14)

bIo

avyabhicarita

rni chad pa dbI (5:5)

dbyibs

anuccheda (K5:36)

rni 'jug

samsthana (5:50)

dbye ba

na vartate (K5:9)

rni rtag

bheda (1289a 8)

'ba' zhig

anitya (2:7)

rni 'phen pa

kevaIa (5: 15; J291a8)

'bras

anaksepa (V5:4)

rni slu ba

karya (5: 14)

'brei pa sarnbandha (2:8,10,20; 5:9,9,12,34; J281a6 , 288b6 , 311a 1) -samgata (2: 10)

avisamvada (2:5)

mig gi gzung ba caksusatva (2:7)

rning can -akhya (J324a 1)

'brei pa sla ba sarnbandhasaukarya (K5:35)

rned asat (2:5)

sbyor ba sam yogin (K2:20)

nasti (2: 10) vina (5:5)

rnedpa

MA rna nges pa anaikanta (5: 11; 1284a4)

rna rnthong adma (5:31; J307a4, 308b5) adrsti (5:34; J3lOb5)

nastita (2: 5)

mod kyang yadyapi (V2:20)

drnigs kyis bkar apoddhara (K5:46; J323b6)

drnigs pa (rna) yin rna 'phangs pa analc)epa (k5: 4)

(na) upaIabhyate (5:50)

Tibetan-Sanskrit Lexicon rmongs . jac,la (V5:14; J289b 1)

bzlog pa nivrtti (V5:36)

TSA-TSHA-DZA

yang dag pa'i don du bhiitarthena (5:4)

YA gtso bo rna yin tsam upasarjana (K5:4) tshad rna pramiiI.la (5: 1;50) tshig vakya (2:5) pacta (5:46; 1323b6) tshig sdud samasa (J281a8) tshogs pa samudliya (K5:50) mtshan nyid lak~aJ:Ia (5:36) mtshungs tulya (K2:5) samanya (V2:5) samya (5:26; 1303a8) rdzas dravya (5:2,4,35)

yang dag 'phrod pa sarilyogin (V2:20) yan lag avayava (K5:50) yid ches apta (2:5) yod sadbhava (2:5) yod pa sat (5:4,4,35) satta (K5:9) utkar~a (V5:36) yod pa nyid satta (J281a6) yod pa'i sgra sacchabda (5:4) yon tan gUQa (5:2,40) yon tan can gUQin (V5:2)

ZHA-ZA gzhan anyat (2:9) gzhan nyid anyatva (5:25)

yon tan gyis phan 'dogs gUQopakara (K5:5; J278a8)

RA-LA

gzban du anyatha (2:11; K5:35; 1312a3)

rang gi ngo bo sVariipa (5: 4)

gzbaI bya min aprameya (V2:7)

rang don svartha (5:34; 13lOb6)

gzhi mthun pa nyid samanadhikaraJ:Iya (5:2,4)

rang dbang med asvatantratva (5:4; 13 13a8)

329

330

Tibetan-Sanskrit Lexicon

rangbzhin

shin tu stong

rupa (5:40)

rab gyur

sutucchaka (V5:40)

shes pa

prakarsa (J278a 8)

rab tu grub

jiieya (K5:35)

bshad

prasiddhi (V5:2)

rab tu 'jug

-akhya (K5:46)

sa las gyur

pravartate (V5:50) pravrtti (V5:50)

parthiva (K5:35)

sa las byung rigs

parthiva (V5:35) jati (5:4,11; J283b3 , 314a5)

sel ba rigs kyi chos

apohya (V5: 14) -nut (V5:31) vyudasa (5:31)

jatidharrna (5:36)

rigs sgra jatisabda (5:4)

sel ba nyid vyudasa (J307a4)

rigs pa yukta (V5:31)

sel bar byed apohate (5:28)

rim bzlog las pratilomyatas (V5:35)

so so ba kun la khyab pa pratyekaparisamapti (V5:36)

re re la yong su rdzogs pa pratyekaparisamapti (K5:36)

so sor rjod byed min aprthakSruti (V5:2)

lam patha (5:40)

so sor snang pratibha: (V5:46)

Ius can milrttatva (2:7)

so sor spobs pratibha (J324a1)

Ius can min amilrta (V2:7)

gsal bar byed pa dyotayati (K5:1)

Ius min amilrta (K2:7)

gsal bya jiieya (V5:35)

len (mi) byed (na) upatta (K5:26)

gsog sutucchaka (K5:40)

SHA-SA shing vrk~a

(5:35)

gsaI -nut (K5:31)

Tibetan-Sanskrit Lexicon bsaI ba yin niralqta (J313b 6) bsal bya apohya (K5:14; 1289a8)

bsal zin niralqta (V5:8)

331

Appendix C Text key to translations of PramiifJasamuccaya In order to make the translations of the Pramiil)asamuccaya that occur in this book more easy to read, it was decided not to encumber the English translation with numerous references to the Tibefan translations from which it was made. But in order to make the English translation more usable to those who wish to compare it with the original texts, those references have been included in this appendix. For each numbered passage that occurs in the English translation, information is given concerning the corresponding folio numbers and line numbers of the Tibetan editions. The key to how that information is presented appears in the following chart. KP

PramiilJasamuccayavrtti. Translated into Tibetan by Gser gyi go cha [= Kanakavarmanl and Dad pa shes rab. Text 5702 in Tibetan Tripilaka: Peking Edition. (Ditmaga 1957b ed.)

VD

PramiilJasamuccayavrtti. Translated into Tibetan by Vasudhararaksita and Seng rgyal. Text 4204 in Tibetan Tripi{aka, Sde dge edition. (Diimaga 1981 ed.)

VP

PramiilJasamuccayavrtti. Translated into Tibetan by Vasudhararak;;ita and Seng rgyal. Text 5701 in Tibetan Tripi(aka: Peking Edition. (Diimaga 1957a ed.)

C.l Pramli/Jasamuccaya Chapter Two: "Svarthanumanapariccheda"

1.1.0 = KPI09a 1, VD27a5 , VP27b6

3.1.0 =KP109b4 , VD27b6, VP28a7

2.1.0 = KP109a5, VD27bl, VP28a2

3.2.0 =KP109B7, VD28a1 , VP28b3

2.3.0 = KP109b2 , VD27b4 , VP28a7

3.4.0 = KPllOa5, VD28a5, VP29a1

332

Text key to translations 4.1.0 = KPllOa7 , VD28a7 , VP29a4

13.1.0"; KP1l2b7 , VD30b2 , VP31b3

4.2.0 = KPllOb2 , VD28b2 , VP29a6

14.0.0 = KPl13a2, VD30b5 , VP31b6

5.1.0 = KPllla3 , VD29a2 , VP29b7

15.0.0

5.2.0 = KPllla6 , VD29a4 , VP30a1

16.0.0 = KP1l3a8 , VD31a2 , VP32a3

=KPlllb 1, VD29a7 , VP30a5

17.0.0 = KP1l3b3 , VD31a5 , VP32a6

6.2.0 = KPlllb3 , VD29b 1, VP30a7

18.0.0 = KP1l3b4 , VD31a6 , VP32a7

7.0.0 = KPlllb7 , VD29b4 , VP30b3

20.0.0 =KP113b7 , VD31b 1, VP32b2

8.1.0 = KP1l1b 8, VD29b6 , VP30b3

21.0.0 = KP114a 1, VD31b3 , VP32b5

9.1.0 = KP1l2a3 , VD29b7 , VP30b6

22.0.0

10.1.0 = KP1l2a5 , VD30a2 , VP31a3

23.0.0 = KP1l4a7 , VD31b7 , VP33a2

11.1.0 =KP1l2b2 , VD30a6 , VP31a6

24.0.0 = KPl14b2 , VD32a2 , VP33a5

6.1.0

333

=KPl13a4, VD30b7 , VP31b7

=KP1l4a5 , VD31b6 , VP32b8

12.0.0 =KP112b6 , VD30b2 , VP31b2

C.2 Pramiil}asamuccaya Chapter Five: "Anyapohaparlk1;la" 8.1.0 = KP157b6 , VD67b7 , VP72a2 1.0.0 = KP156a4, VD66a7 , VP70a8 2.1.0 = KP156a6, VD66b2, VP70b1 2.2.0

= KP156b 1, VD66b3, VP70b4

2.3.0 = KP156b 1, VD66b4, VP70b5 3.0.0 = KP156b3 , VD66b7 , VP70b 8 4.1.0 = KP156b6 , VD67a1, VP71a3 4.3.0 = KP157a2 , VD67a4, VP71a6 4.5.0

=KP157a3 , VD67a5 , VP71a8

5.0.0 = KP157a5, VD67a7 , VP71b 1 6.1.0 =KP157b 1, VD67b2 , VP71b4 7.2.0 = KP157b5, VD67b5, VP71b 8

9.1.0

=KP158a2 , VD68a3 , VP72a4

10.1.0 = KP158a5 , VD68aS, VP72a8 10.2.0 = KP158a7 , VD68a6 , VP72b 1 10.3.0 =KP158b2 , VD68a7 , VP72b3 11.2.0 = KP158b7 , VD68b4 , VP72b 8 11.3.0 = KP159a2 , VD68b6 , VP73a3 12.0.0 = KP159a3 , VD69a1, VP73a4 14.1.0 = KP159a4, VD69a2 , VP73a5 15.1.0 =KP159b 1, VD69a5 , VP73b 1 16.1.0 = KP159b3 , VD69a7 , VP73b 4

334

Text key to translations

17.1.0 = KP159b 8, VD69b3 , VP73b 8

28.1.0 = KP163b3 , VD72b7 ; VP77b2

18.1.0 = KP160a5, VD69b7, VP74a4

29.1.0 = KPl64a2 , VD73a5 , VP77b5

18.2.0 = KP160b3 , VD70a4 , VP74b 1

30.1.0 =KPl64a6 , VD73b 1, VP78a5

18.3.0 = KP160b 8, VD70b 1, VP74b6

33.1.0 = KP165a 1, VD74a2 , VP78b8

19.1.0 = KP161a3 , VD70b 3 , VP75a1

34.1.0

=KP165a4, VD74a4 , VP79a4

20.1.0

= KP161a7, VD70b6, VP75a6

35.0.0 = KP165B 1 , VD74b2 , VP79b 1

25.0.0

= KP163a5, VD72b 1, VP77a2

36.1.0 = KP165b3 , VD74b3 , VP79b4

26.0.0 = KP163b2 , VD72b6 , VP77a8

Selected Bibliography

Ajdukiewicz, Kazimierz 1974 Pragmatic logic. Translated from the Polish Logika pragmatyczna by Olgierd Wojtasiewicz. Dordrecht: D. Reidel Publishing Company. Anacker, Stefan 1984 tr Seven works ofVasubandhu: the Buddhist psychological doctor. Religions of Asia Series, 4. Delhi: Motlial Banarsidass. Andersen, Dines and Helmer Smith 1913 ed The Sutta-Nipata. London: The Pali Text Society. Aung, Shwe Zan and C.A.F. Rhys Davids 1915 tr Points of controversy or subjects of discourse: a translation of the Kathiivatthufrom the Abhidharma-pi{aka. London: The Pali Text Society, 1979 (reprint). Bhartrhari 1963 ed

Vakyapadfya ofBhartrhari. Chapter 3, part 1. Edited with the commentary of Helaraja by K.A. Subramania lyer. Deccan College Monograph Series, 21. Pune: Deccan College.

1965 ed

Vakyapadfya of Bhartrhari. Edited by K.V. Abhyankar and V.P. Limaye. University of Poona Sanskrit and Prakrit Series, 2. Pune: University of Poona.

1977 ed

Bhartrharis Vakyapadfya: Die Mulakarikiis. Edited by Wilhelm Rau. Abhandlung fUr die Kunde des Morgenlandes 42, 4. Wiesbaden: Deutsche MorgenHindische Gesellschaft.

Bhattacharya, Bishnupada 1962 A study in language and meaning: a critical examination of some aspects of Indian semantics. Calcutta: Progressive Publishers. Bhattacharya, Kamaleswar 1971 The dialectical method of Nagarjuna (Translation of the Vigrahavyavartanf from the original Sanskrit with introduction and notes). Journal of Indian Philosophy 1:217-261. Bhattaputra JayamiSra. See KumariIabhatta, 1946 ed.

335

Bibliography

336

Biardeau, Madeleine 1964 tr Bhartrhari Vakyapadfya BrahmaktilJlja: avec la vrtti de Harivroabha. Paris: Editions E. de Boccard. BocheIiski, J.M. 1956 Formale Logic. Munich: Verlag Karl Alber, 1970 (3rd edition). Broad, C.D. 1923 The theory of sensa. Reprinted in PerceiVing, sensing and knowing: a book of readings from twentieth-century sources in the philosophy of perception. Edited, with an introduction, by Robert J. Swartz. Berkeley: University of California Press, 1965. pp. 85-129. Brough, John 1951 Theories of general linguistics in the Sanskrit grammarians. Transactions of the Philological Society, pp. 27-46. 1952

Audumbarayal)a's theory oflanguage. Bulletin of the School of Oriental and African Studies 14:73-77.

1953

Some Indian theories of meaning. Transactions of the Philological Society, pp.161-176.

Bu-ston. See Obermiller, 1932 tr. Buddhaghosa 1931 The Sumangala-viliisinf: Buddhaghosa' s commentary on the Dfghanikiiya. Part 2 (Suttas 8-20). Edited by W. Stede. London: Pali Text Society. Burnyeat, Myles 1980 Can the skeptic live his skepticism? Reprinted in The skeptical tradition. Edited by Myles Burnyeat. Berkeley: University of California Press, 1983. pp.117-148. Butchvarov, Panayot 1966 Resemblance and identity: an examination of the problem of universals. Bloomington and London: Indiana University Press. 1970

The concept of knowledge.

Evansto~:

Northwestern University Press.

Cardona, George 1967-68 Anvaya and vyatireka in Indian grammar. Adyar Library Bulletin 31-32:313-352. 1976

PalJ.ini: a survery of research. Trends in linguistics, state-of-the-art reports, edited by W. Winter. The Hague and Paris: Mouton.

Chatterji, Durgacharan 1929-30 A note on thePramdlJ.asamuccaya. Annals of the Bhandarkar Oriental Research Institute 11:195-196. "

Bibliography Chi,R.S.Y. 1969

337

Buddhistformallogic. London: The Royal Asiatic Society of Great Britain and Ireland.

Ching, Julia 1984 Paradigms of the self in Buddhism and Christianity. Buddhist-Christian Studies 4:31-50. Cohen, Morris R. and Ernest Nagel 1962 An introduction to logic. New York: Harcourt Brace Jovanovich. Come, Edward 1962 Buddhist thought in India: three phases of Buddhist philosophy. Ann Arbor, Michigan: The University of Michigan Press, 1967 (Ann Arbor Paperbacks edition).

1980

Contradictiqns in Buddhist thought Indianisme et bouddhisme: melanges offerts a Mgr. Etienne Lamotte. Publications de L'Institut Orientaliste de Louvain, 23. Louvain-La-Neuve: Universite Catholique de Louvain. Pp. 41-52.

Cooper, Lane 1941 tr Euthyphro. The collected dialogues of Plato, including the letters. Edited by Edith Hamilton and Huntington Cairns. Bollingen Series, 71. Princeton, NJ.: Princeton University Press, 1961. . Copleston, Frederick, S.J. 1946a A history of philosophy. Vol. 1: Greece and Rome. Part 1. Garden City, NY: Image Books, 1962 (reprint). . 1946b

A history ofphilosophy. Vol. 1: Greece and Rome. Part 2. Garden City, NY: Image Books, 1962 (reprint).

Davids, T.W. Rhys 1890 tr The Questions of King Milinda. Sacred Books of the East, edited by F. Max Miiller~ vol. 35. Delhi: Motilal Banarsidass, 1982 (reprint). 1899 tr

Dialogues of the Buddha: part 1. Sacred Books of the Buddhists, edited by F. Max Miiller, vol. 2. London: Luzac & Company, Ltd., 1969 (reprint).

Davids, T.W. Rhys and J. Estlin Carpenter 1890 ed The Dlgha Nikaya. Vol. 1. London: Pali Text Society, 1967 (reprint). Davids, T.W. Rhys and Hermann Oldenberg 1885 tr Vinaya texts. Part 1: The Patimokkha and The Mahavagga I-IV. Sacred Books of the East, edited by F. Max Miiller, vol. 13. Delhi: Motilal Banarsidass, 1982 (reprint). Dharmakirti 1953 ed PramalJavarttika. Edited, with Prajnakaragupta's Vartiktilamkara, by Rahula SaiI1qtyayana. Tibetan Sanskrit Works Series, 1. Patna: Kashi Prasad Jayaswal Research Institute.

338 1960ed

Bibliography The Pramiir.taviirttikam ofDharmaklrti: the first chapter with the auto commentary. Edited by Raniero Gnoli. Serie Orientale Roma, edited by Giuseppe Tucci, vol. 23. Rome: Istituto Italiano per il Medio ed Estremo Oriente.

de Jong, J.W. 1986 Review of Jeffrey Hopkins's Meditation on emptiness. Journal of the International Association ofBuddhist Studies 9(1): 124-128. Dmniiga 1957aed

1957bed

Pramiir.tasamu~cayavrtti.

Translated into ribetan by Vasudhararakl/ita and Seng rgyal. Text 5701 in Tibetan Tripi(aka: Peking Edition, kept in the library of the Otani University. Edited by Daisetz T. Suzuki. Tokyo and Kyoto: Tibetan TripitakaResearch Institute.

Pramiir.tasamuccayavrtti. Translated into Tibetan by Gser gyi go cha and Dad pa shes rab. Text 5702 in Tibetan Tripi(aka: Peking Edition, kept in the library of the Otani University. Edited by Daisetz T. Suzuki. Tokyo and Kyoto: Tibetan Tripitaka Research Institute.

1981 ed

Pramiir.tasamuccayavrtti. Translated into Tibetan by Vasudhararakl/ita and Seng rgyal. Text 4204 in Deruge-han: Chibetto Daizokyo. Ronsobu. Inmeibu 1. [Tibetan Tripi(aka, Sde dge edition, Bstan fJgyur. preserved at the Faculty of Letters, University of Tokyo.] Complied at edited by K. Hayashima, J. Takasaki, Z. Yamaguchi and Y. Ejima. Tokyo: Sekai Seiten KaDko Kyokai.

1982

The Pramiir.tasamuccayavrtti ofDigniiga. Chapter Five: Anyiipohaparuqii. Tibetan text edited, with Jinendrabuddhi' s commentary, by Masaaki Hattori. Memoirs of the Faculty of Letters, no. 21. Kyoto: Kyoto University.

Dravid, Raja Ram

1972

The problem of universals in Indian philosophy. Delhi: MotiIal Banarsidass.

FausboIl, V. The Sutta-nipiita. Sacred Books of the East, edited by F. Max Miiller, vol. 1881 tr 10, part 2. Bound with Muller (1881 tr). Fowler, Harold North 1914 tr Plato. In twelve volumes. Vol 1: .E;uthyphro, Apology, Crito, Phaedo, Phaedrus. The Loeb Classical Library. Cambridge, Mass.: Harvard University Press, 1982 (reprint). Frauwallner, Erich 1929 Bemerkungen zu den Fragmenten Digniigas. Wiener Zeitschriftfiir die

Kunde des Morgenlandes 36:136-139. 1930

Beitritge zur Apohalehre. Part 1, DharmakIrti. Wiener Zeitschriftfor die "

Kunde des Morgenlandes 37:259-283.

Bibliography

339

1932

Beitrage zur Apohalehre. Part 2, translation of DharrnakIrti, first half. Wiener Zeitschriftfiir die Kunde des Morgen/andes 39:247-285.

1933

Beitrage zur Apohalehre. Part 3, translation of DharrnakIrti; conclusion. Wiener Zeitschriftfiir die Kunde des Morgen/andes 40:51-94.

1935

Beitrage zur Apohalehre. Part 4, summary of DharrnakIrti. Wiener Zeitschriftfur die Kunde des Morgen/andes 42:93-102.

1937

Beitrage zur Apohalehre. Part 5, Dharrnottara. Wiener Zeitschriftfur die Kunde des Morgen/andes 44:233-287.

1951

On the date of the Buddhist master of the law Vasubandhu. Serie Orientale Roma, edited by Giuseppe Tucci, vol. 3. Rome: Istituto Italiano per il Medio ed Estremo Oriente.

1953

Geschichte der indischen Philosophie. Vol. 1. Salzburg: Otto MUller Verlag.

1956

Geschichte der indischen Philosophie: Vol. 2. Salzburg: Otto MUller Verlag.

1957

Vasubandhu's Vadavidhib. Wiener Zeitschriftfur die Kunde Sud- und Ostasiens 1:104-146.

1958

Die Philosophie des Buddhismus. Philosophische Studientexte: Texte der indischen Philosophie, edited by Walter Ruben, Vol. 2. Berlin: AkademieVerlag, 1969 (3rd revised edition).

1959a

Zur Erkenntnislehre des klassischen Samkhya-Systems. Wiener Zeitschrift fur die Kunde Sud- und Ostasiens 2:84-139.

1959b

Dignaga, sein Werk und seine Entwicklung. Wiener Zeitschriftfiir die Kunde Sad- und Ostasiens 3:83-164.

1961

Landmarks in the history of Indian logic. Wiener Zeitschriftfur die Kunde Sad- und Ostasiens 5:125-148.

Gautama 1967 ed

Nyayadarsana. Edited, with Bhti~aof Vatsyayana, the Varttika of Uddyotakara, the Ttitparya{fkii ofVacaspati and theParisuddhi of Udayana, by Anantalal Thakur. Mithila Institute Series Ancient Text no. 20. Muzaffarpur: Prakrit Jain Institute.

Gillon, Brendan S. 1986 DharrnakIrti and his theory of inference. Buddhist logic and epistemology: studies in the Buddhist analysis of inference and language. Edited by Bimal Krishna Matilal and Robert D. Evans. Dordrecht and Boston: D. Reidel Publishing Company. pp. 77-87.

340

Bibliography

Gillon, Brendan S. and Richard P. Hayes 1982 The role of the particle eva in (logical) quantification in Sanskrit. Wiener Zeitschriftfur die Kunde Sadasiens 26:195-203. Gillon, Brendan S. and Martha Lile Love· 1980 Indian logic revisited: NyayapraveSa reviewed. Journal of Indian philosophy 8:349-384. Gilson, Etienne . 1955 History of Christian philosophy in the middle ages. London: Sheed and Ward, 1985 (reprint of 1980 edition). 'Gos lo-tsa-ba Gzhon-nu-dpal. See Roerich, 1949-53. Haack, Susan 1978 Philosophy of logics. Cambridge: Cambridge University Press. Halbfass, Wilhelm 1980 The vaise~ika concept of gUJ:13 and the problem of universals. Wiener Zeitschrift fur die Kunde Sadasiens 24:225-238. Hall, Bruce Cameron 1986 The meaning of vijnapti in Vasubandhu's concept of mind. Journal of the International Association ofBuddhist Studies 9(1):7-23. Hallie, Philip P. 1967a Pyrrho. Encyclopedia of Philosophy Vol. 7. Paul Edwards, Editor in chief. New York: Macmillan Publishing. 1967b

Sextus Empiricus. Encyclopedia of Philosophy Vol. 7. Paul Edwards, Editor in chief. New York: Macmillan Publishing.

Hattori Masaaki 1968 tr Dignaga, on perception: being the Pratyalqapariccheda ofDignaga' s Pramal)asamuccaya. Harvard Oriental Series, edited by Daniel H. H. Ingalls, vol. 47. Cambridge, Mass.: Harvard University Press. 1973

Mimarnsaslokavarttika, apohavada-shO no kenkyii. Part 1. Memoirs of the Faculty of Letters, no. 14. Kyoto: Kyoto University.

1975

Mimarnsaslokavarttika, apohavada-shO no kenkyii. Part 2. Memoirs of the Faculty of Letters, no. 15. Kyoto: Kyoto University.

1977

The Sautrantika background of the apoha theory. Buddhist thought and Asian civilization: essays in honour of Herbert V. Guenther on his sixtieth birthday. Emeryville, California: Dharma Press.

1980

Apoha andpratibhti. Sanskrit and Indian studies Festschrift in honor of Daniel H. H. Ingalls, edited by M. Nagatomi, B.K. Matilal, I.M. Masson, and E. Dimock. Dordrecht and Boston: D. Reidel Publishing ~ompany.

Bibliography

341

Hayes, Richard P. 1980· Ditmaga' s views on reasoning (sviirthiinumiina.) Journal of Indian Philosophy 8:219-277 1983

Iinendrabuddhi. Journal of the American Oriental Society 103:709-717.

1984

The question of doctrinalism in the Buddhist epistemologists. Journal of the American Academy ofReligion 52:645-670.

1986a

An interpretation of anyiipoha in Ditmaga's general theory of inference. Buddhist logic and epistemology: studies in the Buddhist analysis of inference and language. Edited by Bimal Krishna Matilal and Robert D. Evans. Dordrecht and Boston: D. Reidel Publishing Company. Pp. 31-57.

1986b

Uddyotakara on the whole and its parts. Tetsugaku--The Journal of Hiroshima Philosophical Society 38:117-29.

1986c

Review of Amar Singh's The heart of Buddhist philosophy: Diimiiga and DharmakTrti. Journal of the International Association ofBuddhist Studies 9(2): 166-172.

Herzberger, Hans G. 1975 Double negation in Buddhist logic. Journal of Indian Philosophy. 3:3-16. 1986

Three systems of Buddhist logic. Buddhist logic and epistemology: studies in the Buddhist analysis of inference and language. Edited by Bimal Krishna Matilal and Robert D. Evans. Dordrecht and Boston: D. Reidel Publishing Company. pp. 59-75.

Herzberger, Radhika 1986 Bhartrhari and the Buddhists: an essay in the development offifth and sixth century Indian thought. Studies of Classical India, 8. Dordrecht: D. Reidel Publishing Company. Hirst, R.J. 1959 1967

The problems of perception. London: George Allen & Unwin Ltd.

Sensa. Encyclopedia of Philosophy Vol. 7. Paul Edwards, Editor in chief. New York: Macmillan Publishing.

Hope, Richard 1952 tr Aristotle: Metaphysics. Ann Arbor: The University of Michigan Press, 1968 (Ann Arbor Paperbacks edition). Horner, LB. 1954 tr The collection of the middle length sayings (Majjhima-nikiiya). Vol 1, The fIrst fifty discourses (mi1lapaJ)J)asa). Pali Text Society Translation Series, 29. London: Pali Text Society. 1957 tr

The collection of the middle length sayings (Majjhima-nikiiya). Vol 2, The middle fIfty discourses (maiihimap3J)J)asa). Pali Text Society Translation Series, 30. London: Pali Text Society.

Bibliography

342 1959 tr

The collection of the middle length sayings (Majjhima-nikaya)."Yol 3, The final fifty discourses (uparipaI}J).asa). Pali Text Society Translation Series, 31. London: Pali Text Society.

1964 tr

Milinda's questions. Sacred Books of the Buddhists, 22. London: The Pali Text Society.

Hospers, John 1953 An introduction to philosophical analysis. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1967 (2nd edition]. ISvarakrsna 1967 ed' .. Samkhya-ktirika. Edited, with the commentary ascribed to Yacaspati Misra Yuktidfpika, by Ram Chandra Pandeya. Delhi: Motilal Banarsidass. Jayawickrama, N.A. 1979 ed Kathtivatthuppakarar;a-aUhakathti. Pall Text Society text series, 169. London: The Pali Text Society. Jhalak"ikar, Bhirnacarya 1874 ed Nyayakosa or dictionary of technical terms of Indian philosophy. Revised and re-edited by Yasudev Shastri Abhyankar. Bombay Sanskrit and Prakrit Series, 49. Pune: Bhandarkar Oriental Research Institute, 1976 (4th edition). Jinendrabuddhi. 1957 Vistilamalavatf PramafJasamuccayatfkti. Translated into Tibetan by BIo gros brtan pa. Text 5766 in Tibetan Tripitaka: Peking Edition, kept in the library of the Otani University. Edited by Daisetz T. Suzuki. Tokyo and Kyoto: Tibetan TripitakaResearch Institute. Kajiyarna Yuichi 1973 Three kinds of affirmation and two kinds of negation in Buddhist philosophy. Wiener Zeitschriftfilr die Kunde Sudasiens 17:161-175. Kalupahana, David J. 1986 Nagtirjuna: the philosophy of the middle way. Albany, N. Y.: State University of New York Press. Karnalasila. See 5antaraksita, 1968 ed. Kanada 196'1 ed

Vaise$ikasutra. Edited with the commentary of Candrananda by Muni 5rf Jambuvijaya. Gaekwad's Oriental Series, 136. Baroda: Oriental Institute.

Kassapa, Bikkhu JagadIsa 1961 ed The Kathtivatthu. Nalanda-devanagarI-PaIi-ganthamaIa. Bihar, India: Bihararajakiya PalipakasanamaI}Qala.

Bibliography

343

Katsura Shoryu 1975 New Sanskrit fragments of the PramalJasamuccaya. Journal of Indian Philosophy 3:67-78. 1976

On Abhidharmakosa VIA. IndogakuhO 3 Nos. 1&2, pp.67-78.

1977

In-mei-shO-ri-mon-ron kenkyii. Part 1. Hiroshima Daigaku Bungakubu Kiyo 37:106-126.

1978

In-mei-sho-ri-mon-ron kenkyii. Part 2. Hiroshima Daigaku Bungakubu Kiyo 38:110-130.

1979a

In-mei-shO-ri-mon-ron kenkyii. Part 3. Hiroshima Daigaku Bungakubu Kiyo 39:63-82.

1979b

The apoha theory of Dignaga. Indogaku Bukkyogaku Kenkyu 28(1):493-489 [sic].

1981

In-mei-sho-ri-mon-ron kenkyii. Part 4. Hiroshima Daigaku Bungakubu Kiyo 41:62-82.

1982

In-mei-shO-ri-mon-ron kenkyii. Part 5. Hiroshima Daigaku Bungakubu Kiyo 42:82-99.

1983

Dignaga on trairiipya. Journal ofIndian and Buddhist Studies 32(1):(15)-(21).

1984

In-mei-shO-ri-mon-ron kenkyii. Part 6. Hiroshima Daigaku Bungakubu Kiyo 44:43-74.

1986a

On trairupya formulae. Buddhism and its Relation to Other Religions: Essays in Honour ofDr. Shozen Kumoi On His Seventieth Birthday. Pp. 161-172. .

1986b

Indo ronrigaku ni okeru henju-gainen no seisei to hatten: Charakasanhita kara Darumakfruti made. [The origin and development of the concept of Vyapti in Indian logic: from the Carakasamhita up to DharmakIrti.J (In Japanese with English abstract pp. 120-122) Hiroshima Daigaku Bungakubu Kiyo, 45. Special edition l. Hiroshima: Hiroshima Daigaku Bungakubu.

1986c

On the origin and development of the concept of vyapti in Indian logic. Tetsugaku--The Journal of Hiroshima Philosophical Society 38: 1-16.

Kitagawa Hidenori 1965 Indo koten ronrigaku no kenlcyu: Jinna (Dignaga) no taikei. Tokyo: Suzuki Gakujutsu Zaidan, 1973 (revised second edition). Knitter, Paul F. 1985 No other name?: a critical survey of Christian attitudes toward the world religions. American Society of Missiology Series, 7. Maryknoll, New York: Orbis Books.

344

Bibliography

Kochumuttom, Thomas A. 1982 A Buddhist doctrine of experience: a new translation and interpretation of the works ofVasubandhu the Yogacarin. Delhi: Motilal Banarsidass. Krishan, Y. 1984 1986

Buddhism and belief in atma. Journal of the International Association of Buddhist Studies 7(2): 117-135. Buddhism and the caste system. Journal of the International Association of Buddhist Studies 9(1):71-83.

Kumarila Bhatta 1898 ed Mfmamsaslokavartikam. Edited with Parthasarathi Misra's Nyayaratntikaravyakhya by Rama S"astri Tailanga. Chowkhamba Sansrit Series, 11. Varanasi: Chowkhamba Sanskrit Series. 1946 ed

Mfmamsaslokavarttikam. Edited with Bhattaputra Jayamisra's Slokavartikatlka (Sarkarika) by C. Kunhan Raja. Madras University Sanskrit Series, 17. Madras: Madras University.

Kunjunni Raja, K. 1963 Indian theories of meaning. Adyar Library Series, 91. Madras: Adyar Library and Research Centre. Kunst, Arnold 1947 tr Kamalaslla's commentary on S"antarak~ita's AnumanaparIk~a of the Tattvasmigraha. Melanges chinois et bouddhiques 8:106-216. La Vallee Poussin, Louis de 1923 tr L'Abhidharmakosa de Vasubandhu. 6 volumes. Second edition with preface by Etienne Lamotte. Melanges chinois et bouddhiques, 16.. Brussels: Institut Beige des Hautes Etudes Chinoises, 1971 (2nd edition). 1937

Documents d'abhidharma: Les deux, les quatre, les trois verires. Melanges chino is et bouddhiques 5: 159-187. Brussels: Institut BeIge des Hautes Etudes Chinoises.

Lamotte, Etienne 1958 Histoire du bouddhisme indien: des origines l' ere Saka. Bibliotheque du Museon, 43. Louvain: Institut Orientaliste, 1967 (reprint).

a

Lewis, C.I. 1944

1946

The modes of meaning. Reprinted from Philosophy and Phenomenological Research, vol. 4, in Semantics and the philosophy of language, edited by Leonard Linksy. Urbana, Chicago and London: Umiversity of Illinois Press, 1952. Pp. 50-63.

An analysis of knowledge and valuation. La Salle, Illinois: Open Court.

Loux, Michaell. " 1974 Ockham's theory of terms: Part I of the Summa Logicae. Notre Dame: University of Notre Dame Press.

Bibliography

345

l'nkasiewicz, Jan 1951 Aristotle's syllogistic: from the standpoint of modern formal logic. Oxford: Oxford University Press, 1957 (2nd edition enlarged). ~adhava, Vidyara~ya

1906 ed

Sarvadarsanasamgraha. Edited, with Madhusiidana Sarasvati's Prasthanabheda, by the pandits of Anandasrama. Ananda5rama Sanskrit Text Series, 5LPune: Ananda5rama.

~aIlavadin

1966 ed

Dvadasaram N ayacakram ofAcarya Srz M allavadi K~amasramalJa. Part 2. Edited with Sirhhasiiri's NyayagamanusiirilJlby Muni Jambiivijaya. 5n Atmanand Jaih GranthamaIa Serial, no. 94. Bhavnagar: 5n Jain Atmanand Sabha.

1976 ed

~alvania,

1967 ~atilal,

Ksamasramana Dviidasaram Nayacakram ofAcarya Sri Mallavadi KsamasramalJa. Part l. Edited with Sirhhasiiri's NyayagamanusarilJlby Muni Jambiivijaya. 5n Atmanand Jain GranthamaIa Serial, no. 92. Bhavnagar: 5n Jain Atmanand Sabha.

Dalsukh Review of D.N. Shastri's Critique of Indian realism. In Journal of the Oriental Institute, M.S. University of Baroda, 16:389-391.

1966

Bimal Krishna Indian theorists on the nature of the sentence (vakya). Foundations of language 2:377-393.

1968

Difmaga's remark on the concept of anumeya. The Journal of the Ganganatha Jha Research Institute, 24: 151-159.

1971

Epistemology, logic and grammar in Indian philosophical analysis. The Hague: Mouton.

1977

Nyaya-VaiSesika. A history ofIndian literature, edited by Jan Gonda, vol. 4, fasc. 2. Wiesbaden: Otto Harrassowitz.

1986a

Perception: ,an essay on classical Indian theories of knowledge. Oxford: Oxford University Press.

1986b

Buddhist logic and epistemology. Buddhist logic and epistemology: studies in the Buddhist analysis of inference and language. Edited by Bimal Krishna Matilal and Robert D. Evans. Dordrecht: 1). Reidel Publishing Company. Pp. 1-30.

~atthews,

1974

P.H. Morphology: an introduction to the theory of word structure. Cambridge Textbooks in Linguistics,!. Cambridge: Cambridge University Press.

346

Bibliography

Mookerjee, Satkari 1935 The Buddhist philosophy of universal flux: an exposition of the philosophy of the critical realism as expounded by the school of Digniiga. Delhi: Motilal Banarsidass, 1975 (reprint). Morris, Richard 1885 ed The Anguttara-Nikiiya. Part 1: ekanipata, dukanipata and tikanipata. London: Pall Text Society. Miiller, F. Max 1881 tr The Dharnmapada: a collection of verses. Sacred Books oithe East, edited by F. Max Miiller, vol. 10, part 1. Delhi: Motilal Banarsidass, 1980 (reprint). Nagarjuna 1951 ed 1960 ed

The Vigrahavyiivartanf. Edited by E.H. Johnston and A. Kunst. Melanges chinois et bouddhiques 9: 102-152. Niigiirjunfyarh Madhyamakasiistrarn. Edited, with CandrakIrti's Prasannapadiivyakhyii, by P.L. Vaidya. Bauddha-sarhslqta-granthavalI, 10. Darbhanga: Mithila-vidyapI\ha.

Norman, K.R. 1984 The group of discourses (Sutta-nipiita). With alternative translations by LB. Homer and Walpola Rahula. Pali Text Society Translation Series, 44. London: Pall Text Society. Oberhammer, Gerhard 1963 Ein Beitrag zu den Vadatraditionen Indiens. Wiener Zeitschriftfii.r die Kunde Sild- und Ostasiens 7:63-104. Obermiller, E. 1932 tr History of Buddhism (Chos ~byung) by Bu-ston. Materialien zur Kunde des Buddhismus, 19. Heidelberg: O. Harrossowitz. Olden berg, Hermann 1929 ed The vinaya pi(aka. Vol 1: The Mahavagga. London: Pali Text Society. Oliver, Curtis F. 1978 Perception in early Nyaya. Journal of Indian Philosophy 6:243-266. Otto, Rudolf 1923 The idea of the holy: an inquiry into the non-rationalfactor in the idea of the divine and its relation to the rational. Translated from Das Heilige by John W. Harvey. London: Oxford University Press, 1958 (paperb~ck edition of 1950 second edition). Patanjali 1880 ed

Vyiikarar;arnahiibhiiD'a. Vol. 1. Edited by F. Kielhorn, revised by K.V. Abhyankar. Pune: Bhandarkar Oriental Research Institute, 1962 (third ed~~. •

Bibliography

347

Popkin, Richard H. 1967· Skepticism. Encyclopedia of Philosophy Vol. 7. Paul Edwards, Editor in chief. New York: Macmillan Publishing. Potter, Karl H. 1963 Presuppositions of India's philosophies. Westport, Connecticut: Greenwood Press Publishers, 1976 (paperback edition of 1972 reprint). 1977 ed

Encyclopedia of Indian philosophies: Indian metaphysics and epistemology. Princeton, New Jersey: Princeton University Press.

PraSastapada 1971 ed Prasastapadabhii~am. Edited, with the commentary KirarJlivalf of Udayanacarya, by Jitendra S. JetJey. Gaekwad's Oriental Series, 154. Baroda: Oriental Institute. Prior,A.N. 1955

Formal logic. Oxford: Oxford University Press, 1962 (2nd edition).

Rahula, Walpola 1959 What the Buddha taught. With a foreword by Paul DemieviIIe. Bedford: Gordon Fraser, 1967 (2nd revised edition). Randle, H.N. 1926 Fragmentsfrom Dinnaga. First edition: The Royal Asiatic Society, London. Delhi: Motilal Banarsidass, 1981 (reprint).

1930

Indian logic in the early schools. London: Oxford University Press.

Rangaswami Iyengar, H.R. 1927 Kumanla and Dignaga.lndian Historical Quarterly 3:603-606. 1951

Bhartrhari and Diimaga. Journal of the Bombay Branch of the Royal Asiatic Society (New Series) Vol. 26, part 2, pp. 147-149.

Reale, Giovanni 1985 A history of ancient philosophy. Vol 3: The systems of the Hellenistic age. Translation by John R. Catan of Storia dellafilosofia antica. 111. 1 sistemi dell' eta ellenistica, published in 5 volumes in 1980. Albany: State University of New York Press. Renou, Louis 1957 Terminologie grammaticale du Sanskrit. Bibliotheque de I'Ecole des Hautes Etudes, no. 280, 281, 282. Paris: Librairie Ancienne Honore Champion. Rescher, Nicholas 1980 Induction: an essay on the justification of inductive reasoning. Oxford: Basil Blackwell.

348

Bibliography

Roerich, G.N. 1949 tr The blue annals. Royal Asiatic Society of Bengal Mongraph Series, 7. Calcutta: Royal Asiatic Society of Bengal. Ross, W.D. 1941 tr

Metaphysica. The basic works of Aristotle. Edited and with an introduction by Richard McKeon. New York: Random House.

Ruegg, David Seyf~rt 1959 Contributions al' histoire de la philosophie linguistique indienne. Paris: Publications de l'Institut de Civilisation Indienne. Samkara. 1980

Siintarak~ita

1968 ed

Brahmasutrasamkarabhi4Yam. Edited, with Govindananda'sBhii$)laratnaprabha, Vacaspatimisra's B hiimati, and Anandagiri 's Nyayanir~aya, by JagadIsa Sastri. Delhi: Motilal Banarsidass. Tattvasangraha. Edited with Kamalas~I1a's Pafijika by Dviirikiidas SasttL 2 volumes. Varanasi: Bauddha BhiiratL

Sasaki Genjun H. 1986 Linguistic approach to Buddhist thought. Delhi: Motilal Banarsidass. Sastri, N. Aiyaswami 1942 tr Ditlnaga'sAlambanaparik$a and Vrtti. Restored with the commentary of Dharmapala into Sanskrit from the Tibetan and Chinese versions and edited with English translations and notes with extracts from Vinltadeva's commentary. Madras: The AdyarLibrary. Sharma, Dhirendra 1969 The differentiation theory of meaning in Indian logic. The Hague: Mouton. Shastri, Dharmendra Nath 1964 Critique of Indian realism: a study of the conflict between the Nyayavaise~ika and the Buddhist Dignaga school. Agra: Agra University. Shorey, Paul 1930 tr Republic. The collected dialogues of Plato, including the letters. Edited by Edith Hamilton and Huntington Cairns. Bollingen Series, 71. Princeton, N.J.: Princeton University Press, 1961. Singh,Amar 1984 The heart of Buddhist philosophy: Dinniiga and Dharmakirti. New Delhi: Munshiram Manoharlal. Sprung, Mervyn 1979 tr Lucid exposition of the middle way: the essential chapters from the Prasannapada of Candrakfrti. Translated in collaboration with T.R. V. Murti and U.S. Vyas. Boulder, Colorado: Prajfia Press.

Bibliography Staal, J.F. 1962

349

Contraposition in Indian logic. Logic, methodology and philosophy of science:proceedings of the 1960 international congress. Stanford.

1965

Reification, quotation and nominalization. Contributions to logic and methodology in honor of I.M. Bochenski. Edited by A.-T. Tymieniecka. Amsterdam: North-Holland Publishing Co. pp. 151-187.

1967

Indian logic. Encyclopedia of Philosophy Vol. 4. Paul Edwards, Editor in chief. New York: Macmillan Publishing.

Stcherbatsky, Th. 1923 The central conception of Buddhism and the meaning of the word "dharma." Delhi: Motilal Banarsidass, 1970 (4th ed.). 1930

Buddhist logic. Vol. 2. Bibliotheca Buddhica, 26. Leningrad: Izdatel'stov Akademii Nauk S.S.S.R.

1932

Buddhist logic. Vol. 1. Bibliotheca Buddhica, 26. Leningrad: Izdatel'stov Akademii Nauk S.S.S.R.

Steinkellner, Ernst 1971 Wirklichkeit und Begriffbei DharmakIrti. Wiener Zeitschriftfur die Kunde SUdasiens 15: 179-212. 1982

The spiritual place of the epistemological. tradition in Buddhism. Nanto Bukkyo (Journal of the Nanto Society for Buddhist Studies) 49:1-15.

Streng, Frederick J. 1967 Emptiness: a study in religious meaning. Nashville: Abingdon Press. Subramania Iyer, K.A. 1969 Bhartrhari: a study of the Vakyapadfya in the light of the ancient commentaries. Pune: Deccan College. 1971 tr

The Vakyapadfya ofBhartrhari. Chapter 3, part 1. Deccan College Building Centenary and Silver Jubilee Series, 7l. Pune: Deccan College.

Suzuki, Daisetz Teitaro 1953 Essays in Zen Buddhism: third series. Edited by Christmas Humphreys. London: Rider and Company, 1985 (reprint). Tachikawa Musashi 1970-72 A sixth-century manual of Indian logic (a translation of the NyayapraveSa). Journal of Indian Philosophy 1:111-145. Tillemans, Tom J.F. 1986 Dharmaldrti, Aryadeva and Dharmapala on scriptural authority. Tetsugaku--The Journal of Hiroshima Philosophical Society 38:31-47.

Bibliography

350

Tola, Fernando and Carmen Dragonetti 1982 Dignaga 's Alambanaparlk$iivrtti. Journal ofIndian Philosophy 10: 105-134. Toulmin, Stephen Edelston 1958 The uses of argument. Cambridge: Cambridge University Press, 1980 (reprint of 1964 paperback edition). Tredennick, Hugh 1933 tr Aristotle. In twenty-three volumes. Vol 17: The Metaphysics, Books 1-9. The Loeb Classical Library. Cambridge, Mass.: Harvard University Press, 1980 (reprint). 1935 tr

Aristotle. In twenty-three volumes. Vol 18: The Metaphysics, Books 10-14. Bound with Oeconomica and Magna Moraiia, translated by G. Cyril Armstrong. The Loeb Classical Library. Cambridge, Mass.: Harvard University Press, 1977 (reprint).

1954 tr

Socrates' defense (apology). The collected dialogues of Plato, including the letters. Edited by Edith Hamilton and Huntington Cairns. Bollingen Series, 71. Princeton, N.J.: Princeton University Press, 1961.

Trenckner, V. 1880 ed The Milindapanho, being dialogues between King Milinda and the Buddhist sage Niigasena. London: Paii Text Society, 1962 (reprint). 1887 ed

The Majjhima-Nikiiya. Vol. 1. London: Pali Text Society.

Tripathi, Chhote Lal 1972 The problem of knowledge in YogiiciiraBuddhism. Varanasi: BharatBharati. 1975

The role of apoha in Dignaga's theory of knowledge. East and West 25:455-470.

Tucci, Giuseppe 1928 On the fragments from Dignaga. Journal of the Royal Asiatic Society: 377-390. 1929

Buddhist logic before Diimaga. Journal of the Royal Asiatic Society: 451-488.

Udayana. See PraSastapada 1971 ed. Uddyotakara Bhliradvaja 1916 ed Nyiiyaviirttika. Edited by V.P. Dvivedin and L.S. Dravid. Kashi Sanskrit Series, 33. Varanasi: Kashi Sanskrit Series. Vaidya, P.L. Saddharmalankiivatiirasiitram. Buddhist Sanskrit Texts, 3.Darbhanga: 1963 ed MithilavidyapItha. .

Bibliography

351

Va mana and Jayaditya 1967 ed Kasikavrtti. Edited with the Nyasaparaparyayakasikavivaraf}apaficika of Iinendrabuddhi and Padamafijari of Haradattamisra by Dvarikadasa Sastri and Kalikaprasada Sukla. Prachya Bharati Series, no. 5. Varanasi: Tara Publications. van def Kuijp, L.W.J. 1979 Introductory.notes to the Pramaf}avilrttika based on Tibetan sources. The Tibet journal 4(2):6-28. Vasubandhu 1967 ed AbhidharmakosabhaD'am ofVasubandhu. Edited by P. Praclhan, revised by Arona Haldar. Tibetan Sanskrit Works Series, 8. Patna: K.P. Jayasawal Research Institute, 1975 (2nd revised edition). 1970 ed

Abhidharmakosam. Kosasthana 1-2. Edited, with Vasubandhu's BhaD'a and Yasomitra's Sphu{ilrtha{fka, by SvlimI Dvarikadiis 5astrI. BauddhabharatigranthamaIa 5. Varanasi: Baudclha Bharati.

1971 ed

Abhidharmakosam. Kosasthana 3-4. Edited, with Vasubandhu's Bha:;ya and Yasomitra's Sphu{artha{fka, by SvlimI Dvarikiidas 5astrI. BauddhabharatigranthamaIa 6. Varanasi: Bauddha Bharati.

1972 ed

Abhidharmakosam. Kosasthana 5-6. Edited, with Vasubandhu's BhaD'a and Yasomitra's Sphu{artha{lkii, by SvlimI Dvarikadiis 5astrI. BauddhabharatigranthamaIa 7. Varanasi: Bauddha Bharati.

1973 ed

Abhidharmakosam. Kosasthana 7-8. Edited, with Vasubandhu's BhaD'a and Yasomitra's Sphu{artha{lkii, by SvlimI Dvarikadas 5astrI. BauddhabharatigranthamaIa 5. Varanasi: Bauddha Bharati.

Venn,John 1881 1907

Symbolic logic. New York: Burt Franklin, 1971 (reprint of 2nd revised edition of 1894). The principles of empirical or inductive logic. New York: Burt Franklin, 1972 (reprint).

Vidyabhii~aI;la,

1905

Satis Chandra Dignaga and his PramilI;la-samuccaya. Journal and proceedings of the Asiatic Society ofBengal, new series vol. 1, pp. 217-27.

1907a

Indian logic as preserved in Tibet Journal and proceedings of the Asiatic Society ofBengal, new series vol. 3, pp. 95-102, 241-255, 541-551.

1907b

Sanskrit works on literature, grammar, rhetoric and a lexicography as preserved in Tibet. Journal and proceedings of the Asiatic Society of Bengal, new series vol. 3, pp. 121-132.

1907c

Nyaya-praveSa, or the earliest work extant on Buddhist logic by Dignaga. Journal and proceedings of the Asiatic Society of Bengal, new series vol. 3, pp. 609-617.

352

Bibliography

1907d

Hetu-cakra-hamaru or Dignaga's Wheel of Reasons--recovered from Labrang in Sikkim. Journal and proceedings of the Asiatic Society of Bengal, new series vol. 3, pp. 609-617.

1909

History of the mediaeval school ofIndian logic. New Delhi: OritfntaI Books Reprint Corporation, 1977 (2nd edition).

1921

A history ofIndian logic: ancient, mediaeval and modern schools. Delhi: Motilal Banarsidass, 1971 (reprint).

Warder, A.K. 1970 Indian Buddhism. Delhi: Motilal Banarsidass. 1971

Outline ofIndian philosophy. Delhi: Motilal Banarsidass.

Wayman, Alex . 1958 The rules of debate according to Asanga. Journal of the American Oriental Society 78:29-40. 1979

Yogacara and the Buddhist logicians. Journal of the International Association ofBuddhist Studies 2(?):65-78.

Woodward, F.L. 1960 tr The book of the gradual sayings (Anguttara-nikaya): or more-numbered sayings. Vol 1 (Ones, two, threes). With an introduction by Mrs Rhys Davids. London: Pall Text Society. Wylie, Turrell 1959 A standard system of Tibetan transcription. Harvard journal ofAsiatic studies 22:261-267. Yamaguchi Susumu 1929 tr Dignaga' s Examen de I' objet de la connaissance [Alambanaparf~a]: textes tiMtain et chinois et traduction des stances et du commentaire eclaircissements et notes d' apres Ie commentaire tibetain de Vinftadeva. In collaboration with Henriette Meyer. Paris: Imprimerie Nationale.

Subject and Author Index anvaya: 119 anyiipoha: 158 aphasia, Skeptic philosophers': 52 apoha theory of meaning: 8, 13, 14, 18-21,26-29,202,210,278; Dharmottara's: 22; Diilnaga's: 23 ApohaprakarafJ.am: 21 applicability (vrtti; 'jug pal: 255 apposition: 194, 253 appositional phrase: 198 apprehension of oneself: 84 Aquinas, St Thomas: 77 archetype (arkhe): 82; Aristotle's: 74; Plato's: 74; substance and: 75 arguable property (sadhya): 146 arhant: 97 Aristoc1es the Paripatetic: 51 Aristotle: 74, 146,258; Ockham's appeal to: 77 art of persuasion: 93 arthiik~epa: 276 Asailga: 6, 12, 171 association (anvaya): 118,120, 130, 151,241,298; in grammar: 190 ataraxia, Skeptic philosophers': 52 Ansa: 226 atom: 96; (paramiinu): 103, 104 atomic proposition: 281 audible symbols: 189 Averroes, Ockham's appeal to: 77 avijjii: 42 awareness (vijfiiina; vififiiifJ.a): 81, 86 ayoniso manasikiira: 42, 63

A Abhidhamma Pitaka: 92 abhidbarma: xii; 6,23,28,73, 88, 93, 94,96,100,142,143,170,173, 227,233 Abhidharmakosa: 29, 94,103,173 Absolute, transcendental intuition of: 89 absolute simple: 96, 174 absolutely singular: 75, 96 abstract notions: 105 abstract nouns: 30 abstract property (dharma): 100 abstraction: 211 abstraction from others (anyiipoha): 95 abstraction suffixes: 270 accidentally (shugs kyis; *siimiirthyiit): 294 action (karman): 206 action (laiya): 206,271 adjectival symbol, nature of: 197 adjectival term: 206 adjectives: degrees of comparison: 275; (gufJ.asabda): 203 afflictions (kida): 53 Agama, Buddhist: 42, 50 agent (kartr): 271 Ajdukiewicz, Kazimierz: 112 Aiambanaparllqii: 132, 173 . Alailkadeva: 10 Alexander (commentator on Aristotle): 146 Alexander the Great: 51 ambiguity: 297 analogy (upamiina): 61, 70 anger, impairment of judgement due to: 46 Ailguttara Nikaya: 45 "anonymous Christian," Rahner's concept of: 34

B Bahuleya: 273 Barbara syllogism: 130 Bhamaha: 19 Bhartrhari: ix, 27, 28,30,31, 178-182, 251,256-261,270,278,280,297,

353

354 299,311; DiI'maga influenced by: 13,23,29,30,41,212,238 Bhattaputra J ayamisra: 223, 224, 262, 263,265,272,302,304,305,307, 308 Blo gros brtan pa: 228 Blue Annals: 227 Boo Ie, George: 169 Brahmajala Sutta: 45 brahman: 43, 49, 50, 51, 78, 80 Brahmanism: 23 Broad, C.D.: 134 'Brog mi: 226 Bstan 'gyur: 9,223,224; Sde dge (Derge) edition: 25,228 Bu ston Rin po che: 4,226,229 Buddha Gautama (Gotama): 42,46,50, 79,81,82,84,90; caste, views on: 79; KaIamas, advice to: 48; Socrates compared with: 49 Buddha's words (buddhavacana): 33; authority of: 5 Buddhaghosa: 107 Buddhism: Canonical: 35; schools of 17 Buddhist practice: 6 Buddhist Sangha: 79 Burnyeat, Myles: 52 Butchvarov, Panayot: 274

C calm (upasama): 81 Candragomin: 229 Candrananda:249 Candrarahula: 226 Caraka: 29 Cardona, George: 119,302 Carvaka: 44 case affixes (vibhakti): 171 (See also inflectional affixes.) caste (jali) system: 79 catalepsy (abhisanna-niTodha): 82 categories: 235 causal efficiency (arthakriya): 14, 15, 18,20,23,26,225 causal relation: 18,24, 163,254,260 causation: 143 character (sarizsktira; sarizkMra): 81, 86 Chatterji, Durgacharan: 10 Chenna: 10 Ching, Julia: 106 Citta Hatthisariputto: 84

Index class, generated by language: 210 class containment: 113 class disjunction: 113 class overlapping: 113 cognition (jnana): 94 colour (Tapa): 100,235 common-sense: 16 common usage (racjhi; grags pay: 237 complex entity (samudaya; bsdus pa): 282 conceit (mana): 85 concept: 14,15, 18, 19,23,24,37; (vikalpa; rnam par rtog paY: 238

configuration (akrti): 19 confirmation (sadhana): 111 connotation: 19; Satkari Mookerjee's usage of term: 19 containment (class relation): 113 contaminants (asrava): 94 contemplation (dhyana; jhana): 66, 83 context (prakaraTJa): 295 contradiction, law of; Nagarjuna's adherence to: 58 contrariety of terms: 206; indirect: 207; Vaise~ika view: 207 contrary terms: 205 conventional truth: 94 conventions, for usage of symbols: 19 Conze, Edward: 4,5, 88 co-reference: 259,260,264,274,278, 283; (samanadhikaraTJya): 187, 262,278 counterextension: 158, 254 counterfactuals: 164 creator: 46 criterion (pramaTJa): 60 criterion ofreality: 14 ncri tical realism n: 16 Cynic school of philosophy: 51

D Dad pa'i shes rab: 228 de Jong, J.W.: 36 death, existence after: 45,90 debate: 93 defining attribute (lak~ar;a): 56, 73 deliverance (apavarga): 253 demonstration: 29; (parartMnumana): 25,26 denotation, Satkari Mookerjee's usage of term: 19

Index derivative idea: 54,88,91,92 desire (ir"TJa; taTJhti): 85 Dge lugs pa school: 4 . Dharmadharmatavibhmigakiirikii: 227

DharmakIrti: x, xiii, 3-5, 7,10, 11, 14-17,20-22,24,29,30,32,33,35, 37,95,119,162,171,224,253, 260,308-310; Ditmaga contrasted with: 15,20, 137; inference according to: 252; sensation according to: 137; Stcherbatsky's view of as "moderate": 12 Dharmottara:x,4, 7,10,11,17,21,22 differentiation (bheda): 27 dialectics: 89,90,91 DIgha Nikaya: 45 Ditmaga: ix; xi, 1,2,4-17,20,22-31, 78,95,111-127,135,136,149, 150, 153-158, 163, 166, 175, 178-183,186-190,193-205, 208-215,223,226,229,235,236, 239,240,246,253-269,272-289, 294,297,303,308-311; association according to: 196; DharmakIrti contrasted with: 15,20, 137; dissociation according to: 196; evidence, 3 characteristics of according to: 121; inference according to: 254; linguistic signs according to: 188; mind, view of: 138; Nyaya school, critique of: 136, 206; particulars according to: 189; reasoning according to: 180; semantic theory of: 2597; Stcherbatsky's view of as antiMMhyamaka, anti-skeptic and "moderate" 12, 13; truth, theory of: 204; universals, argument against: 182; Vasubandhu's influence on: 172 Diogenes Laertius: 51 discipline (sikkhti): 82 disjunction (of classes): 113 disregarding the differences (bhedagraha): 28

disjunctive syllogism: 276 dissociation (vyatireka): 118, 120, 130, 152,158,241,298; in grammar: 190 distress (dub/cha; dukkha): 50,81, 84 divinity (devata): 253 dogmatics, Buddhist 33

355 domain (vi~aya): 193 doxastic minimalism: 52 Dravid, Raja Ram: 15,20,32,246 dravya: 27

E earth (prthivJ): 235 eidos, to: Plato's: 73; Socrates's: 73 element (bhuta): 56; (dhtitu): 100 emancipation: 5 (See also nirval)a.)

empirical knowledge: 88 empirical realm: 81 Encompassment (iik~epa; 'phen pay relation: 261 epistemology: 1, 16,23,29,32; Buddhism, place of in: 3,6 emptiness (sunyatii): 56,61,71,93; as antidote to opinions: 62 epiphenomenon: 90 episteme, in Plato; 74 errancy (vyabhiciira; 'khrul paY: 113, 155,255 essence (svabhiiva): 95 essential aspect (to eidos): 73, 86 essential man (puru~a) in Samkhya: 264 essential relationship (svabhtivapratibandha): 162

ether (iikiisa): 235, 277 Euthyphro: 72 evidence (hetu): 113, 146, 181,231, 253; legitimate, three characteristics (trirupa) of: 121 excluded middle, law of; Nagarjuna's adherence to: 58 exclusion (apoha): 184,211 expressor relation (viicyaviicakasambandha): 260

extension: 169, 254; of linguistic sign: 193,278 extensionally (arthatas): 294 external object (bahyiirtha): 132 ex ternal reali ty: 19 external sandhi: 297

F faculty ofreason: 216 faith: 5 fallacy (jiiti): 25 feelings (vedana): 81, 86

356 field of operation (vi:)aya; yul): 174, 232 figurative language: 199; constraints on: 265 fire (tejas): 235 folk designation (loka-vohiira): 85 folk expression (loka-nirutti): 85 folk idea (loka-pafdiatti): 85 folk name (loka-samafifia): 85 foolish masses of people (Mlaputhujjana): 110 formal debate: 130; (viida): 145 Frauwallner, Erich: 10, 16,20-24, 29-32,37,41,108,133,163,277, 305,310 full awakening (sambodhi): 81

G gamogenesis: 90 Gailgesa: 10 general attribute (samanyalak:)alJa): 135,238 general term: 195-199,205,261,288; (jatisabda; rigs kyi sgra): 202, 255, 257; relative scope of: 288 genitive case: 146 genus-species relations: 163,287 Gillon, Brendan S.: xiv, 172 Gnoli, Raniero: 38 gnosis: 34 gnt5sis, in Plato: 74 God: 34 'Gos khug pa: 226 'Gos 10 tsa ba Gzhon nu dpal: 227 "Gotama's razor": 50 grammar (vyakaralJa): 229 grammarians, Sanskrit: 119, 156, 190, 200,206,250,275,280,302,304; Difmaga influenced by: 41 grammatical case: 146 grammatical subordination (upasarjana): 262 Grand Path (Mahayana): 96 grounds of application (pravrttinimitta): 206, 207 (See also warrant for application.) gymnosophists, influence on Skeptics: 51

Index H Haack, Susan: 250 habit-energy (vasana): 32 Hall, Bruce Cameron: 98,99, 109 "hand-maiden of theology": 5 happiness, true (siva): 53, 55 Haradatta Misra: 305 Hattori Masaaki: 26-31,36,38,132, 139,170,226,230,302,306-308 Hayes, Elizabeth A.: xiv, 39 Hayes, Richard P.: 36-39, 109, 159, 170,250 "Hayes-Ditmaga" system of logic, H.G. Herzberger's critique of: 159 head word (pradhiina): 262 heaven (svarga): 239, 253 Helaraja: 260,278,305 hell: 239 Hellenistic Skeptics: xiii Herzberger, Hans G.: 157 Herzberger, Radhika: 26, 30-32, 38,41, 311 Hetucakranir7}aya: 111-131, 145, 151-154,159,161,165,241,242 hetu-trainlpya: 121 homomorphs: 297 homonymity: 296 Hospers, John: 135 Humble Path (Hfnayana): 96 hypothesis: 112; (pratijfia): 60

I ideal form (to eidos); 104 Idealists: 13 identity: 54, 78, 87 identity relation: 194 identity theory of universals: 274 idiomatic expression (ru4hi; grags paY: 271 impact (pratighata), three types of: 173 incompatibility (virodha); 113 Indian logic: 29, 30 indirect implication (arthak$epa); 276 individuals: 32,75; (vyakti); 19,255, 261 induction domain: 113, 118, 125, 127, 129,242

Index inerrancy (avyabhicara): 255 . inferable object (anumeya): 144 inference: 1,5, 14, 15,20,23,29, 178-180,253; fallibility of: 178, 180; test for accuracy of: 14; theory of, Diimaga's: 24; theory of, N yaya-Vaise~ika: 24 inferential sign (/inga).· 8,24,29, l32, 146,181,187,203,211,245,255, 277,294 inflectional affixes: 216 inner peace (ajjhattasantim), by abandoning opinions: 44, 45 insight (cakkhu): 81, 84 insight (vipa§yana) meditation and logic: 168 instantiation of universal (jatimat): 196, 261,263,269 instrumental cause (karal'Ja): 141 intension: 169,197; linguistic sign, of: 193,278 intentional object (dngos po; *vastu): 259 intensionally (sabdatas): 294 intellect: 68, 74, 82, 105 interpretation of signs: 1, 33 intuition (pratibhti): 28 Isipatana (Deer Park): 81 Isvarasena: 224

J

Jainas: 17 Jainism: 79 JambUvijaya, Muni: 304,307 jati: meaning of term: 106 JhalakIkar, Bhlmacarya: 261 jhtina, first: 66 Jayanta Bhatta: 20 Jinendrabuddhi: xiv, 8, 11, 12,26,199, 200,223-225,228-234,238,243, 245,250,251,253,255,258,259, 262,264,265,268-270,273-275, 284,285,290-296,298,302, 304-306 Jfianasrimitra: x, 7 Jfianendra:305 Joyce, James: 89 judgement: 14, 15,37, 135,255; impairment by emotion of: 46; test for accuracy of: 23

357

K Kaiyata: 305 Kalamas: 48 ka/panti: 135 KamalasIla: x, 5, 7, 10, 16,18,27,170, 253,263,300,303,306-308 Kanakavarman: 25,219,224,234,247, 248,257,264,272,296,302,304 Kapila: 253 karman: 79, 251 (See a/so action.) Kasikavrtti: 229 Kathavatthu: 91 Katsura Shoryu: x, xiii, 10,29,251 Katyayana: 30, 31 Kim Samu Sunim: xiv Kitagawa Hidenori: 25, 26, 29, 31, 39, 132,238,243 Knitter, Paul F.: 40 Kochumuttom, Thomas A.: 98, 109 Krishan, Y.: 79, 106 Kumarila Bhatta: xiii, 8,10, 16, 17, 19, 22,27,199,270,303 Kun dga' rgyal mtshan: 3 Kunjunni Raja, K.: 27

L La Vallee Poussin, Louis de: 93, 108 /aksana: 90; and to eidos compared: 73 Lamo'tte, Etienne: 79 language: 1, 15,20, 30; Diimaga's theory of: 32, 33 Lankavatara Sutra: 103 lexeme: 261,265,269,288,297,302 linga:26 linguistic convention: 208, 252, 255, 294 linguistic science (sgra rig pa; *sabdavidya): 229 linguistic sign: 187, 203,250-252,277 living being (sattva): 90 locative case: 146 Locke, John: 98 logic: 10,23,29, 163; Buddhism, place of in: 3, 5,12, l3; Dinnaga's theory of: 32, 35; logically complete sentence: 194 Lukasiewicz, Jan: 146, 169

358 M Madhava, VidyaraI).ya: 44 Madhyamaka school: 16,25,88; early phase of: xi, 41, 144; mysticism and: 12,58,61; skepticism and: 12, 13 Magi, influence on Skeptics: 51 Mahayana Buddhism: 3, 12 Mahayana sillras: 6,42 Maitreyanatha: 227 major premiss: 130 Mallavadin KsamasramaI).a: xiii, 8 Malvania, Dalsukh: 7 Manibhadraraksita: 227 Maim, social codes of: 79 material form (nlpa): 95 materialism, Carvaka: 44 Matilal, Bimal Krishna: x, xiii, 171, 203,251,302,306 MATup (taddhitapratyaya): 263, 272 Matthews, P.R.: 302 means of acquiring knowledge (pramalJa): 253 means of understanding (buddhyupaya; rtogs pa'i thabs): 237 meditation: 100 (See also jhana) meditation adept, sensation of (yogipratyak:;a): 136 memory: 273, 274 mental image: 19 mentality (caitta): 175 metaphysical categories: 206 metaphysical doctrines: 20 metaphysical parsimony: 13 metaphysical presuppositions: 20 metaphysical terminology: 204 metaphysics: 10, 16, 163 metonymy (upacara; btags pay: 265 Metrodorus of Chios: 51 Milinda, King: 67, 86 IvITmarnsa school: 10, 17; "naive realism" and: 16; sensation according to: 132 MTmiiltlsaslokavarttika: xiii, 270, 272, 303 mind (manas): 83, 135, 140,235 minor premiss: 130 mirage: 140 misconception (avidya; avijja): 43, 94 modus tollendo ponens: 276

Index Mok~akaragupta: 229 Mookerjee, Satkari: 15-22,24,32,97, 309 moral behaviour, justification for: 90 Mulamadhyamakakarika: 55, 92 muni: 110 mysticism: 34

N Nagarjuna: 13,35,41,52-59,69,70, 88-92, 104, 109, 144 Nagasena: 67, 86 "naive realism": 12, 16 narrower term (visesasabda): 288 nature (prakrti): 266 necessarily (dngos su; sak:;at): 294 Nikaya literature: 42 nirvalJa: 3, 18,33,34,42,69,81,83, 84,86,92,94,100,104,312; Buddhist: 52; differing opinions of: 47; logic and: 168; Nagarjuna's view of: 55; "numinous" nature of: 34 noble eight-fold path: 81 nominal affixes (Sup): 302 nominal compound (samasa; tshig sdud): 270, 281 nominal existence (prajiiapti-sat): 28 nominal stem (pratipadika): 302 nominalism: 13,28,35,41,92,95, 173; Buddhist: 20; Ditmaga's: 205; objection to: 210; Ockham's: 77, 87 nominalist theory of universals: 274 nominative case: 146 non-empirical realm: 81 numen, Otto's concept of: 34 Nyaya school: 8, 13, 16, 17,23,206, 262,275; Diimaga citicized by: 142; early: 10; inference theory of: 24; modem: 10; "naive realism" and: 12, 16; pratyak$a, definition of: 139; sensation according to: 132 Nyayabindu: 8,11,21 Nyayamukha: 23; Chinese translations: 29; Japanese translation: 29 Nyiiyasutra: 139 Nyayavarttika: xiii Nyi ma rgyal mtshan: 229

Index

o

observed precedent (dr:j!iinta): 25, 237 obverse proposition: 119' Ockham, William: 77,87,96 opinion (dr~ti; di(!hi): 85; abandoning: 52; contamination through: 94; ill consequences of: 44; impediments to inner peace: 44, 51,92; nonBuddhist teachers': 46 Otto, Rudolf: 34 overlapping of classes: 113

p Pa tshab 10 tsa ba Nyi rna grags pa: 228 pak:;adharmatii: 121 PaIi Canon: 42 PilQini: ix, 2, 30 Paramatthakasutta: 43 Parthasarathimisra: 10,223,224,264,

270,272,302-308 particular: 14, 15, 18, 19,21,23,24,28, 73, 192,255,273; (bheda; khyad par): 257, 276,304; (svalaksana):

26,27

..

particular proposition: 154 particularity: 31 particularizing expression (bhedasabda; khyad par kyi sgra): 255 Pataiijali: ix, 27, 272; universals according to: 183 peculiar attribute (svalaksana): 238

273

perc8~tion:

. . , 5; (sarhjiiii; saiiiiii): 81, 82,

perfection of wisdom (prajiiapiiramitii):

89

'

person: 86, 90, 94; as mental construct:

45 pervasion (vyiipti; khyab pa): 29,113,

156,158,256,263 phenomenalism: 93, 131, 138, 173 phenomenon (vijiiapti): 99 physical resistance (iivarafJ£Jpratighata): 174 Pisacas: 253 "point-instant": 14, 15,20 Plato: 73,78,81,96,258; Aristotle's criticisms of: 75 pleasure, impairment of judgement due to: 46

359 pondering (vikalpa): 53 Potthapada: 82, 83 Potthapada Sutta: 82, 92 Prajfiakaragupta: 169-171 prajiiiipiiramitii: 227 pramiilJa: 5; meaning of the term: 70 PramiilJasamuccaya: x, 8, 10, 17,23,

26,27,129,162,172,177,187, 189,192,194,195,198,204-214, 218,219-243,273,277-280,285, 290,298,303, 306-308; English translation: 26; Japanese translation: 25; Tibetan translations: 8,25 PramiilJasamuccaya!fkr'i: 228 PramiilJaviirttika: 21, 22,26, 38, 260 ~8 ' prapaiica: 53, 68 pratibhii:212 preclusion (apoha): 223, 297 preconception: 135 predication: 196; simple: 193 Priestley, L.C.D.C.: 63 primary elements: 207 primary suffix (lqtpratyaya): 271 primitive idea: 92 primordial matter (prakrti): 253 primordial substance (pradhana; gtso boY: 180,238,239,266 principled thinking (yoniso manaskara):

92,93 process of elimination: 235 proof: 112 properties: 1 property-!?earer (dharmin): 137 property-possessor (dharmin): 146 property to be confirmed (siidhyadharma): 113,241,254 proposition: atomic: 281; defined: 56; obverse: 119; particular: 152; singular: 152; universal: 152; universal affirmative: 119,167 psychologism: 240 pudgala; puggala: 87 Pyrrho ofElis: 51,67 Pythagoras: 1

Q qualificand (vise:jya): 286 qualifier (vise:;ar;a): 286

Index

360 qualifier-qualificand relation

(vise~al)a­

vise~yabhava):

31, 187,275,278; symmetrical and asymmetrical: 291 quality (gUlJa): 31, 206, 271 quasi-evidence (hetvabhiisa; gtan tshigs Itar snang baY: 241

R radical momentariness, doctrine of (klalJikavada): 15, 19,23 Rahner, Karl: 40 Rahula, Walpola: 48 Rak~asas:

253 Randle, H.N.: 10,250,251 Rangaswami Iyengar, H. R.: 10,27,

305,308 rational discussion (veida): 60 RatnakIrti: x, 7,16,20 Realists: l3, 20 reality (sattva): 207 reasoning (anumana): 1,61,70, l35 reasoning for oneself (svarthiinumana):

1,23,25,26 reference: S. MookeIjee's usage of term: 19 rejaxepoid: 190 relations: 19, 259; expression of in language: 195, 197; reality of: 13; (samsarga): 27; (sambandha): 270 religious duty (dharma) in Brahmanism:

179 Renou, Louis: 301 representationism: 97 resemblance: 246,273; (sarupya; 'dra bay: 265 resemblance theory of universals: 274 resistance (pratigha; thogs pay: 236 revelation: 6; Buddhism, place of in: 12,

58 rigorous truth (paramarthasat; paramatthasat): 94 Rgyal tshab Dar rna rin chen: 253 Rma 10 tsa ba: 227 Rngog 10 tsa ba: 227 Ruegg, David Seyfort: x, 230, 280 rupadhiitu: 10 1

S Sa skya PaI)<;lita: 3 Sabaleya: 273

Sakyabuddhi: 22 samanadhikaralJya: 286 Sarhkaracarya: 304 Samkhya Karika: 253 S:lrhkhya school: 17,23, 143, 180,239, 253,266; Diimaga influenced by: l3 3; sensation according to: 132 samyagjfiana: 5 sanketa: 256 Sanskrit language: xiii, 2, 146 Santabhadra: 10 Santarak~ita: x, 5, 7,16, 17, 18, 19,27,

255,263,309 Sarvastivada: 16 satisfaction (of a symbol): 214 Sautrantika school: 17,23,27,29,96 saviour (soter): 35 scriptural tradition (agama): 61, 70, 178,236; accountability to reason:

181 schools of Buddhism: 17 Sde dge (Derge) Bstan 'gyur: 25,232 secondary suffix (taddhitapratyaya):

195,263,271 sectarian affiliations, Dinnaga's: 17 seer (Ni): 179 segregation (apoha): 211 self: 91,94; (atman; attan): 53, 54, 78,

81,82,87,235,266 self-cognizing cognition: 140 self-referring expression: 360 semantics: 19 sensation: 14,135, 168, 178-180,255, 274; (pratyak~a): 24, 61, 70, 132, 134,139,234,273,283; yogin's:

l36 sense datum: 100, 134, 176,233,255, 273 (See also sensum) sense faculties: 100, 135,233 sensible objects: 26, 78 sensum/sensa: 135 (See also "sense . datum") sentence: function of words in: 280; lo~ically complete: 193; primary umt of speech: 212; (vakya; ngag): 237,302; meaning of (vakyartha):

28,212,215,238,254,290 sentential homomoIphism: 297 set-subset relation (tadbhedatva): 198 Sextus Empiricus: 52, 68 Sharma, Dhirendra: 15

Index Shastri, Dhannendra Nath: 7, 15,20, 32, 97,309 sign: conventional: 1; linguistic: 211; natural: 1; (linga): 231 simple predication: 194 Singh, Amar: 97, 108 singular proposition: 154 singular term: 188, 189,206; (yadrcchiisabda): 203 "skill in means" (upiiyakausalya): 4 skeptical rationalism: 41, 62 smell (gandha): 100,235 smoke: (See fire.) Socrates: 49, 52, 72, 86 soteriology: 10, 34 soul, reality of: 13 sound sabda: 100 space, as an element: 58; (dis): 235 Stcherbatsky, Th.: 3,11-17,21-22,37, 58,61,97,230 Steinkellner, Ernst: 5, 33 string of words (tshogs; *samuha): 290 structure/structuring (kalpanii): 134, 234,274 stupid chatter (appii{ihfrakatam bhiisitam): 84 subject (palqa) of inference: 113 subject matter (vi~aya): 94 subjective constructions: 19 subjective idealism: 16, 98, 131 subjectlike class (sapalqa): 116, 118 subordinate word: 262 subordination, grammatical (upasarjana): 197,262 substance (dravya): 209, 235 substance (ousia), Aristotle's: 75 superior knowledge (abhUifitj): 46, 81; nature of Tathagata's: 47 support (iilambana): 174 Sutta Nipata: 44, 83, 105, 311 SuttaPitaka: 100 Suzuki, Daisetz Teitaro: 89, 107 symbol: class: 192, 200, 206, 288; meaning of: 19; token: 192,200, 288 synonymous terms: 285,288 synonymy: 289

T tadbhedatva: 198 Taranatha: 10

361 Tarkabhiisii: 228 taste (rasa): 100,235 Tathagata: 46, 85 Tattvasangraha: 8, 18, 303 term: Diimaga's five types of: 203 theology: 10, 163 "thing universal" (arthajiiti): 31 three features (trairupya): 29 Tibetan Tripitaka: 227; Peking edition: 25 time (kala): 235 Ting nge 'dzin bzang po: 226 touch (spada): 100,235 Toulmin, Stephen: 70 traidhiitukam: 101 Traikiilyaparik~ii: 23 transcendental intuition: 89 Tripathi, Chhote Lal: 15 truth: 15; conventional: 94; ultimate: 93 Tsong kha pa Blo bzangs grags pa: 253 Tucci, Giuseppe: 10

U Udayana: 11,249 Uddyotakara: xiii, 8, 10, 16, 19,22,27, 142,150,171,199,223,250,258, 264,265,302-304; Diimaga citicized by: 149 ultimate fact (paramiirtha; paramattha): 88 ultimate truth: 93 understanding (fiiiT}a): 81 unity: 75; relationship with diversity: 85 universal: 13-15,18-20,24,28,30-32, 41,75,85,187,210,246,254,261, 265,270,283; abstraction, as: 211; ontological status of: 19; signification of words, as: 195; Uiiti; rigs): 28, 183,206,270,276; Diimaga's definition of: 296; grammarian's definition of: 271; properties of: 210, 300; relation to particular instantiation: 274; (samiinya; spyi): 32; (siimiinyalaksaJ)a): 26, 27 universal-possessor: 196 universal affirmative proposition: 119; implicit existential import of: 169 universal proposition: 154 unrelated to thought (cittaviprayukta): 28

362 unsubjectlike class 118 Upani~ads: 79 upasarjana: 263

Index (asapak~a):

116,

V vacaka relation: 257 Vacaspatimisra: 10, 11,20 Vaibh~ika school: 25,28,96 Vaise~ika school: 8, 13, 16,23, 143, 206, 235, 236, 262, 282; inference according to: 24; "naive realism" and: 12; sensation according to: 132 Vaise~ika Siltras: 249 Vajapyayana: 27,28,31 Vakyapadfya: 179, 180,217,218,256, 259,260,278,302,306 van der Kuijp, L.W.J.: 4, 5 Vasetthasutta: 79 Vaseiiha and Bharadvaja, dispute on caste: 80 Vasubandhu: 11,12,28,41,92,94-98, 101,103,138,170,173,177,239, 311; Diimaga influenced by: 97, 173; sensation according to: 132 Vasudhararak~ita: 25,220,224,227, 228,238,249-251,259,266,272, 296,303,307 vedana:48 Vedanta school: 17 Vedas: 49 Vedic injunctions: 31 Venka\acarya, T.: xiii Venn, John: 114 verb root (dhtitu): 302 verbal affixes (TIn): 302 verbal communication, as form of inference: 181 verbal conjugational affixes (TIil): 325 verbal symbol (sabda): 211; parallelism with inferential signs: 298 verbal term: 206 verbal testimony (sabda; sgra las byung bay: 237, 253 verbs and deverbati ves (kriyasabda):

203 Vetter, T.: 5,34 via antiqua and via moderna (Aristotelian schools): 77 Vidyabhil~aI)a, Satis Chandra: 9, 10, . 163,251

Vigrahavyavartanf: 52, 59 Vijiiaptimatratasiddhi: 103,239 Vinaya Pij:aka: 91 Vinltadeva: 10,11,217 vipa~a: 254 Visalamalavatf: 228 vise~alJavise~yabhava: 286 visible quality (rapa): 209 vital principle: 90 vrtti relation: 257 Vya~i: 27, 31, 32 VyakaralJamahabhti~a: 305 vyatireka: 119 vyavahtira: 256

W Warder, A.K.: xiii, 6, 32, 79, 309 warrant for application (pravrttinimitta): 261,280 (See also grounds of application.) water (apas): 235 Wayman, Alex: 4, 98 Weiss, Jason: xiv whole-part relation: 194, 195,286; expression of in language: 194 wider term (samanyasabda): 288 wind (vayu): 235 word: individual (pada; tshig): 237, 302;meaning of: 19,215 word form: 265 word token: 265 "word universal" (sabdajati): 31 world of matter (rapaloka): 100

X Xuanzang: 29, 108,253

y Yasomitra: 173 Yijing: 10,25,30 Yogacara school: 16,23,25,96,129, 140 Yuktidfpika: 253

Z Zen masters: 89 Zha rna rdo rje rgyal mtshan: 227

Index Zha rna Seng ge: 227

363

DIGNAGA ON THE INTERPRETATION OF SIGNS

by RICHARD P. HAYES Faculty of Religious Studies, McGill University, Montreal, Quebec , Canada

Beginning with the work of Dignaga in the sixth century, Indian Buddhist scholastics turned their attention increasingly more tot the problems of epistemology and the study of natural language l The impetus to take up these issues was Buddhist opposition to the Brahmanical doctrine of the special authority of revealed scripture, in contrast to which the Buddhist philosophers celebrated reason as the final authority in deciding what is true. Despite the central place that the philosopher Dignaga held in the history of Indian Buddhism, his principal work in this area has not before been thoroughly studied. In this book Hayes traces Dignaga's principal epistemological theories to the Buddhist skeptical tradition represented in such early works as the Sutta Nipata and the philosophy of Nagarjuna; Dignaga's theories on language are described in the context of the philosopher's opposition to the great grammarian Bhartrhari. The second part of the book contains the first English translations of selected passages of Dignaga's principal writings on inference and philosophy of language. ~

KLUWER ACADEMIC PUBLISHERS DORDRECHT / BOSTON / LONDON

ISBN 90-277- 2667- 1